“how did the woman seal my tongue?” Paul can’t tell his father he’s going to die. Why? Sometimes our tongues are bound to speak that which can’t be changed
“the mind commands the body and it obeys. It commands itself and it rebels”
-
Connections: - Alan Pinkerton (killers of the flower moon) “secrets break the character of the criminal till they give it out”. Dr Yiwei cries when his scripture is given out.
-
-
if I made him do it it wouldn’t be his doing.
-
storytelling: why tell us 100% of everything up front? The movie hides the secrets about Yueh.
-
storytelling: the hindsight narrator
-
+
Connections: - Alan Pinkerton killers “secrets break the character of the criminal till they give it out”. Dr Yueh cries when his scripture is given out.
+
Notes: - if I made him do it it wouldn’t be his doing. - storytelling: why tell us 100% of everything up front? The movie hides the secrets about Yueh. - storytelling: the hindsight narrator
When God has appointed a man to die, he guides his wants to desire that place
Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only.
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/docs/index.html b/docs/index.html
index 277e1361c..79edd13bb 100644
--- a/docs/index.html
+++ b/docs/index.html
@@ -119,11 +119,9 @@
-
-
@@ -272,7 +270,7 @@
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/docs/search.json b/docs/search.json
index 9b27587b7..7e17e2982 100644
--- a/docs/search.json
+++ b/docs/search.json
@@ -10,353 +10,360 @@
]
},
{
- "objectID": "gratitude.html",
- "href": "gratitude.html",
- "title": "Gratitude",
+ "objectID": "little-lamb.png.html",
+ "href": "little-lamb.png.html",
+ "title": "Bryan Whiting",
"section": "",
- "text": "“Gratus sum, ergo habeo.” (I’m thankful, therefore I have.)\nIn November 2020, a Christian apostle named Russell Nelson invited the world to flood social media with gratitude. I rather enjoyed posting gratitude publicly, so here we go.\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nI’m grateful for blankets\n\n\nMy grandma made blankets cool\n\n\n\n\n\n\nDec 26, 2023\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nI’m grateful for doctors\n\n\nDoctors of the body. Doctors of the mind. Doctors of the soul…Any engaged in the pursuit of healing.\n\n\n\n\n\n\nDec 25, 2023\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nLight\n\n\nI set up 72 lights today.\n\n\n\n\n\n\nDec 24, 2023\n\n\n\n\n\n\n\n\nNo matching items\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only.",
- "crumbs": [
- "Gratitude"
- ]
+ "text": "_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2017-10-18-used-cars-scrape/used-cars-scrape.html",
- "href": "posts/2017-10-18-used-cars-scrape/used-cars-scrape.html",
- "title": "Buying a used car the data science way: Part 1",
+ "objectID": "posts/2022-06-21-the-power-of-party-pay-by-visible/index.html",
+ "href": "posts/2022-06-21-the-power-of-party-pay-by-visible/index.html",
+ "title": "The Power of Party Pay by Visible",
"section": "",
- "text": "[Update 2021-11-16] This analysis was originally written on my old blog here. You can find the source code for it here. This code no longer works as TrueCar changed their CSS to make it more difficult to scrape. It’s still possible, but you’d need to build a custom scraper from scratch.\nThis is part 1 out of a two-part series on scraping used car data. Check out part 2 to learn how to analyze the data.\nIn another post, I describe how I use this data that I’ve scraped, but I wanted to provide a more in-depth tutorial for those interested in how I got the data. Note, this data belongs to Truecar, so all uses herein are for personal and academic reasons only.\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Visible’s party pay madness\nVisible wireless offers $25 off to users who join a party. So a user who signs up to pay $50 for unlimited data can click one button to join a party and only have to pay $25/mo from then on. Why not just offer users $25 to begin with?\nIt’s really confusing to me as a user - why not just give me the $25 automatically if it takes nothing to join a group? Because of gamification.\n\n\nWhy it works\nI think their strategy is as follows:\n\nThey hope some people get brought in on marketing and never remember to activate a party pay. Let’s say this number is 20-40%. Perhaps these users fund the rest of the discounts.\nJoining a party pay group for free is an easy way to make your customers win. I.e., a user gets $25 off for joining. Gimmicky, but I think it works. So a user has an immediate sense of gratification - a great way to build a strong emotional bond. “Click a button to get $25 off in perpetuity? Nice!” Anyone would agree this feels better than just starting off with $25.\nThis makes users want to share and tell others. Especially if customers feel they’ve gamed the system. Who today doesn’t like a hack?\n\nBut even if (1) is wrong, and that 95% of people join a party I’m sure they’ve done the math to consider how many people they need to sign up to make the money work.\n\n\nThe Larger Strategy\nWhy would Verizon cannibalize their own sales? I switched my wife out of a $70 unlimited plan on Verizon to a $25 plan on visible for the same service. Well some customers will stay with Verizon because of its prestige and the add on benefits. But users who are cutting the cord are going somewhere else, I’m thinking Verizon wants to stay ahead of the curve. It’s a race to the bottom, sure. But it doesn’t hurt to try to win the race.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2017-10-18-used-cars-scrape/used-cars-scrape.html#get-the-data",
- "href": "posts/2017-10-18-used-cars-scrape/used-cars-scrape.html#get-the-data",
- "title": "Buying a used car the data science way: Part 1",
- "section": "Get the data",
- "text": "Get the data\nIn order to do any good analaysis, you first need data. I prefer to have more data than less, where possible. In this case, I don’t have any data, so I use webscraping to get the data. There are much better tutorials on how to scrape data, so I’ll be light. I use R’s rvest package here, which does a decent job.1 Let’s look at Truecar’s Used Car postings2. First I use google to find the search query on Truecar that I like."
+ "objectID": "posts/2022-06-28-pomodoro-principles/index.html",
+ "href": "posts/2022-06-28-pomodoro-principles/index.html",
+ "title": "Pomodoro Principles",
+ "section": "",
+ "text": "I’ve been trying a Pomodoro timer in TickTick and it’s changing how I approach my work day. I’ve been using it for two days so far and here are the differences I notice:\n\nWhen my mind knows I’ll get a break at 25 minutes, it allows me to focus more.\nIt’s easier to say not to slack messages or impulses to open the news if I know I’ll have time for those things later.\nA 5 min break is perfect.\nIf a distraction pops up, I write it down in my journal and tell myself I can follow up during my break.\n25 minutes is short. It’s amazing how time flies.\n\nI used to hate timing myself after working in consulting. I vowed I would never work on the clock ever again. But don’t the best athletes pace themselves? Any marathon trainer knows how to set a pace and have rest periods. Why wouldn’t the same principles apply to cognitive work?\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2017-10-18-used-cars-scrape/used-cars-scrape.html#load-packages",
- "href": "posts/2017-10-18-used-cars-scrape/used-cars-scrape.html#load-packages",
- "title": "Buying a used car the data science way: Part 1",
- "section": "Load packages",
- "text": "Load packages\n\nlibrary(rvest)\nlibrary(dplyr)\nlibrary(magrittr)\n# Find the URL of the data you want to scrape\nurl <- 'https://www.truecar.com/used-cars-for-sale/listings/ford/edge/'\nread_html(url)\n\n## {xml_document}\n## <html lang=\"en-US\">\n## [1] <head>\\n<meta http-equiv=\"Content-Type\" content=\"text/html; charset= ...\n## [2] <body>\\n <!-- Element target for any partner code meant to execut ...\nYou’ll see there’s a head and a body. Our data’s in the body, so let’s use html_nodes() and html_text() to parse out the data we want. I used Selectorgadget to know what HTML classes to search for.\n\nread_html(url) %>% html_nodes('.col-xs-6.col-sm-8.no-left-padding') %>% html_text()\n\n## character(0)\nSo that’s how you get the data on a single page. If you look closer at the URL, you see a lot of helpful things. First, there’s the make, then the model, then the location-zip, then the year-range, and ultimately the trim. This is a very pretty and clean URL. If you click on a few additional pages, you’ll see the URL opens up with ?page=2.\nhttps://www.truecar.com/used-cars-for-sale/listings/ford/edge/location-90210/?page=2\nThis is our ‘in’ to scraping multiple pages. I won’t bore you with the details of how to get that data into a neat matrix for us to analyze, but suffice it to say that I’m able to do it. Just build a function to construct a URL, and build a loop to go through the different pages, then use lots of str_extract() from the stringr package and gsub to clean up the data.\n\nlibrary(stringr)\n\nmake = 'ford'\nmodel = 'edge'\nzip = '90210'\nyear = 2012\nnpages = 5\n\nurl <- paste('https://www.truecar.com/used-cars-for-sale/listings/', \n make, '/', \n model ,\n '/location-', zip,\n '/year-',year,'-max/?page=', sep = \"\")\n\nurls <- paste(url, 1:npages, sep = \"\")\n\nscrape <- function(pageno){\n try(\n read_html(urls[pageno]) %>% html_nodes('.col-xs-6.col-sm-8.no-left-padding') %>% html_text()\n )\n}\n\nlong_list = scrape(1)\nfor(i in 2:npages){\n print(i)\n new_list = try(scrape(i))\n \n error = (\"try-error\" %in% class(new_list))\n \n if( error == FALSE ){\n long_list = c(long_list, new_list) \n } else {\n break\n }\n}\n## [1] 2\nstats <- long_list\ndf <- as.data.frame(stats)\ndf$stats %<>% as.character()\ndf$price <- str_extract(df$stats, '\\\\$[0-9]*,[0-9]*') %>% \n gsub('Price: |\\\\$|,', '', .) %>%\n as.numeric()\ndf$year <- str_extract(df$stats, '^[0-9]* ') %>% \n as.numeric()\ndf$mileage <- str_extract(df$stats, 'Mileage: [0-9]*,[0-9]*') %>% \n gsub('Mileage: |,', '', .) %>%\n as.numeric()\n\n# a = df$stats[1]\ndf$trim <- str_extract(df$stats, '.*Mileage:') %>% \n gsub('FWD|AWD|4x[24]|[24]WD|V6|4-cyl|^[0-9][0-9][0-9][0-9]|4dr|Automatic|Manual|Mileage:', '', ., ignore.case = T) %>% \n gsub(make, '', ., ignore.case = T) %>% \n gsub(model, '', ., ignore.case = T) %>% \n trimws() \n\n\ndf$awd <- grepl('AWD|4WD|4x4', df$stats, ignore.case = T) %>% as.numeric()\ndf$manual <- grepl('manual', df$stats) %>% as.numeric()\ndf$v6 <- grepl('V6', df$stats) %>% as.numeric()\ndf$location <- str_extract(df$stats, 'Location: .*Exterior:') %>% \n gsub('Location: |Exterior:', '', .) %>% \n trimws() \ndf$ext <- str_extract(df$stats, 'Exterior: .*Interior:') %>% \n gsub('Interior:|Exterior:', '', .) %>% \n trimws() \ndf$int <- str_extract(df$stats, 'Interior: .*VIN:') %>% \n gsub('Interior: |VIN:', '', .) %>% \n trimws() \ndf$vin <- str_extract(df$stats, 'VIN: .*\\\\$') %>% \n gsub('VIN: |\\\\$', '', .) %>% \n substr(., 1, 17)\ndf$deal <- str_extract(df$stats, '\\\\$[0-9]*,[0-9]* below') %>% \n gsub('below|\\\\$|,', '', .) %>% trimws() %>%\n as.numeric()\n\nAnd here’s what the results look like. You’ve got the original scraped data in the stats column and then everything else that you can parse out.\n\n# df was the dataframe object we needed\ndf %>% select(-stats) %>% head(10) %>% formattable::formattable()"
+ "objectID": "posts/2022-06-28-increase-focus-through-writing-down-everything/index.html",
+ "href": "posts/2022-06-28-increase-focus-through-writing-down-everything/index.html",
+ "title": "Increase Focus through Writing Down Everything",
+ "section": "",
+ "text": "The GTD framework preaches something to the effect of “get every todo item down and into a system. It may take dozens or hundreds of tasks to get it all out.” Essentially, if a task is in your mind, it will slow you down. But if you put it down on paper, it will free your mind to think on other things. Also, more ideas will come.\nI’ve put perhaps a hundred things down on paper in the last few days. Interestingly, ideas I had forgotten about came back to my mind. I wrote these down.\nThere got to a point today though where I had so much on my screen I felt overwhelmed. How will I ever sort through all this mess?\nThankfully, most of the things don’t need to get done and are either “eventually” or “someday maybe” tasks, and not “asap” tasks. So just ignore them.\nAnd secondly, I always “know” what’s most important. This documentation process doesn’t change that. It just helps me to get all the distractions out of my head and down on paper.\nSpending time organizing all this stuff seems like a potential waste. But what I realize is that it improves my just-in-time recall. If earlier in the day I planned to buy a dish scrubber in the evening, when the evening rolls around that task magically comes back to my mind. So it’s really just front loading all the decision making and prioritizing. I think our brains are capable of more subconscious processing than we realize.\n(Pause entry to buy dish scrubber…) okay, I’m back.\nAnother idea that’s having an effect on me is the 2 min rule. If a task takes less than 2 min, don’t organize it. Just do it. Powerfully, this mindset helps me to care less about two min tasks. Or, sometimes there’s a tendency to over exaggerate the stress something will cause. Like buying a scrubber could be a 10 min task if you want to find the perfect one. But if you only allocate yourself 2 min, you realize this is only a $6 decision and I have better things to do with my time.\nLastly, I love being deliberate about tasks by writing down the intended outcome of a task. It’s very easy as a data scientist to get swallowed up in all the weeds. Or, rather, when your task is to dig up a gem it’s easy to get distracted by all the weeds that need to be pulled that are in your way. So by being clear about the reason I’m diving into the data in the first place helps me re-surface less scathed. Data are a dirty business…\n\nTakeaways\nThe effect of the GTD framework?\n\nLess FOMO. By writing down everything I can see clearly that 1) I don’t have time for all of it and 2) the stuff I care about pops up.\nI don’t have to worry about what I’m not doing because I can easily say no to it as I say yes to something better.\nI can see broader connections between a litany of things. This helps me prioritize what my subconscious has been telling me by nagging me about X ideas.\nIf I start working on something not in my list it makes me ask if this thing is fealty worth my time when I know I have a ton of other stuff to do. (Recency bias I spoke about in my other blog post.)\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2022-06-22-downstream-impact/index.html",
- "href": "posts/2022-06-22-downstream-impact/index.html",
- "title": "Downstream Impact",
+ "objectID": "posts/write-despite-ai/index.html",
+ "href": "posts/write-despite-ai/index.html",
+ "title": "Write despite AI",
"section": "",
- "text": "Every pea is unique. But it’s still a pea.\n\n\nOnline experiments are powerful tools to see the causal impact of a business strategy. Tweak something, see what users do.\nThe limitations of online experiments is that once you manipulate your user experience there’s no going back. Users are forever impacted by your choice.\nIf you give out a credit card with $100, $300, and $500 lines, people are going to spend more on the $500 line. They might also default more because they’re less able to pay back $500 than they are $100. As a result, you may lose these $500 users forever and might impact other aspects of their life, not to mention impact the marketing department’s strategy.\nA group of users is like a single living organism, if you treat it like that. If you only measure one metric (conversion rate) on the entire massive audience, then you’ll optimize for the “average” user.\nBut nobody is average. We are all distinct. We fall into categories, sure. But to really optimize the whole business we need to optimize for pockets of users. Or, we just need to set up a system that’s personalized.\nTakeaway? Personalize your experiments where possible. Consider the ethical and downstream impacts of the experiment.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Write because you can\nI don’t write because anyone is actually listening. But if I’m lucky, the AI is actually listening.\n\nIf AI is listening (webscraping me), I’m fortunate to teach it. Essentially, good training data is hard to come by, especially as AI content proliferates on the internet. Let’s consider it a privilege I’m able to help it learn.\nWriting and telling stories are the oldest form of humanity. Even if an AI can do it better, perhaps we’ll just spend time doing more of it.\n\n\n\n\nWhen plume meets tablet\n\n\nSo yea, I’ll write despite AI. I’ll write because it’s very human to do so.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/quarto-feature-requests/index.html",
- "href": "posts/quarto-feature-requests/index.html",
- "title": "Things I’d like in a blog",
+ "objectID": "posts/2022-12-23-who-do-you-want-to-be-known-for/index.html",
+ "href": "posts/2022-12-23-who-do-you-want-to-be-known-for/index.html",
+ "title": "How do you want to be remembered?",
"section": "",
- "text": "quarto has strengths and drawbacks\n\n\n\nOther frameworks (like Hugo) are more flexible. But quarto offers a lot more out of the box and handles the stuff I really don’t want to have to think about.\n\n\n\n\n\nA traditional Quarto\n\n\nThings quarto is good at:\n\nHandles metadata really well (easy to get a title image, description, Twitter card, etc.)\nheavily customizable: can organize posts as a list, as a table, as a grid, etc. can easily customize what’s in that. The main thing I didn’t like about Hugo was that the style was just slightly different than I liked it.\ncomments\nsearching\ncategories/tags\neasy to create different/custom listings\ncan blog from my phone using Working Copy and GitHub actions\n\nThings wish quarto were better at:\n\nFaster render (Hugo speeds plz)\nDon’t render future posts\nA blog landing page that had a stream of entries, rather than just a list. (Similar to seths.blog)\nListings based on tags\nmake it easy to have an alternate slug (make all posts at root, and not in posts/)\n\nThings I’d like to do with my blog:\n\ngenerate weekly summaries\nbuild some subscribers\nsend out a weekly summary of content.\nauto-post to Twitter, Instagram (using photo + description + zapier) and Facebook.\nauto-post anything with “LinkedIn” tag to LinkedIn\nmove editing into a vault in Obsidian\nmake it easier to copy ChatGPT photos to my blog post folder\n\nFor now, what it offers widely covers what it doesn’t. Most blog templates don’t allow for search, custom listings, lists of tags, Twitter content, etc. For now, quarto offers 95% of what I need.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "On the sports fields of high school, I was often asked, “How do you want to be remembered?” The energy behind that question is a rallying cry for someone to go out and score some goals and be victorious. Score goals → be remembered forever in the annals of high school lore, the thinking goes.\nThis question misses the point.\nIn “Ego is the Enemy”, Ryan Holiday asks the rhetorical question “to be or to do”. The mindset is that either we can focus on doing great work (spending time on action and education), or we can focus on being great (talking too much, dreaming too much). Ironically, greatness follows those who do, not those who say they will do.\nSo instead of asking “how will others remember me”, ask “what will I do for others?”\n\n\n\nDan Kiefer via Unsplash\n\n\nOn this Christmas season, remember that while Jesus Christ’s divinity has been debated for centuries, certainly he’s been remembered. He’s been remembered because of the actions he took and the kindness he showed.\nChristians believe he was the Son of God, one with literal power over death and rose from his own grave. He could calm the seas and turn water to wine. He was the rightful heir to the throne that his Roman captors sat on. He could have easily freed his oppressed nation to rest on his laurels forever.\nYet the recorded miracles we have from him don’t involve him showing moving mountains or collecting power and gold. The reason he’s remembered is because His miracles involve him spending time with individuals and in their service. He healed the sick. Caused the blind to see. He alleviated suffering.\nMay this Christmas be one where you don’t focus on what others think of you. May it be one where you enjoy following His example of doing good for others, however small.\n\nWhosoever will be great among you, let him be your minister; And whosoever will be chief among you, let him be your servant: Even as the Son of man came not to be ministered unto, but to minister, and to give his life a ransom for many. Matthew 20:26–28\n\nMerry Christmas.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2022-06-21-products-for-dads/index.html",
- "href": "posts/2022-06-21-products-for-dads/index.html",
- "title": "Products for dads",
+ "objectID": "posts/writing-is-enlivening/index.html",
+ "href": "posts/writing-is-enlivening/index.html",
+ "title": "Writing is truly enlivening",
"section": "",
- "text": "Products I enjoy as a dad\nThird times a charm - I hope by(/if) kid number four I’ll have this all figured out. It’s amazing how I feel so new to it all despite raising two kids in the last five years. In starting fresh a third time I feel inspired to write down some things I’m rediscovering.\nTo anyone getting started on their baby journey, perhaps these suggestions can help. Your baby is definitely different than mine, so your mileage may vary! But since I’m still using this stuff five years later maybe it can help you too. Either way, congrats and good luck!\n\nProducts I still use on baby 3\nSince I’m a mega influencer (on my very established data science blog here), I get paid $500 every time you click a link below. So please, go crazy.\nI’ll also credit my wife for figuring out to buy these products in the first place.\n\nStretchy swaddle, like this. I like stretchy swaddles because they’re forgiving. The cotton muslin swaddles are nice and soft, but if the fabric doesn’t stretch the right way you may have to redo the swaddle.\nOllie swaddle (nighttime). This thing is ridiculously overpriced for a swaddle. $70? Don’t get it. But if you do, I like it because its super easy to put on and very sturdy (hard for baby to break). At month 2 the baby gets strong and the basic swaddles are so easy to break. That and I feel like I’m gonna hurt my kid if I try to make a cotton swaddle unbreakable. Enter Velcro! I can be gentle on the swaddle but have it be unbreakable. There are a lot of random contraption swaddles that tie the baby up like a pretzel. This is your classic burrito. Insert baby, wrap right side, Velcro left side and you’re done.\nUbbi Steel Odor Locking diaperpail. I also hated the thought five years ago of shelling out $70 for a trash can. Five years since and the thing still performs. I know how well it’s working when garbage day rolls around and I open it to change it. Smells like a gentle coastal breeze, maybe. To this day my wife hasn’t changed it more than a time or two because thankfully I have an iron stomach and unfortunately she does not. Point is: it keeps the diaper smell inside (for the most part) and uses normal garbage bags (cost saver).\nBaby K’tan Original Baby Wrap Carrier, Infant and Child Sling. I’ve tried several carriers. Boba wraps take way too much patience to learn how to put on (patience I ran out of quickly as a new dad). This thing is just two circular fabric swaths, which means it takes two seconds to throw on. The convenience is huge - baby crying? Grab the carrier and snuggle up in 10 seconds or less.\nDockATot Deluxe+ Dock. We just got one off Facebook marketplace for $50 for third kid. We got it because let’s be real - sleeping is hard. The hardest thing about sleeping is the uncertainty what’s happening on the other side of the room in the crib. Grunting. Coughing. Is baby waking up? Needs a pacifier? When it’s 5am and you’re on feeding three of the night, being able to just roll over and plop the little one into the co-sleeper and monitor the situation close up is nice.\n4moms rockaRoo Baby Swing. There are a thousand ways to solve the “where do I put my baby while cooking dinner” problem. We’ve liked this one.\n\n\n\nMust Haves\n\nDish soap. Honestly, nothing gets out a blowout stain like dish soap. That and drying in the sun is a miracle worker. (Apparently this is how cloth diaper people do it.)\nApple cider vinegar for diaper rash. Once the little pimples appear, you’ll try anything to get them to go away. We just discovered this hack recently and I was delightfully surprised how easily it worked.\nThis is the right way to carry a car seat. A friend of mine that has three kids didn’t know this was a thing until he saw me do it. It’s a thing. Do it! Your forearm will thank you.\n\n\n\nToddler Time\nDown the road, these products have hit home runs for me.\n\nMunchkin Miracle 360 Trainer Cup. Every time I put this cup together I consider applying to Munchkin for a job to be their sales rep. This company has very clever designs on their products and this one just crushes it for me. Acts like a normal cup. Performs like a sippy. (I’ll admit I’ve taken a few swigs out of this just to understand how it works!)\nContigo Spill-Proof Kids Tumbler. I think we have 80 water bottles in a junk drawer somewhere. Cheap ones. Expensive ones. They all end up scattered in pieces like the zebra herd before the lion that is your innocent child. This cup is no exception: I’ve lost a straw on one already. But! If you want a water bottle you’ll have the problem of spare parts. (Uhh, this isn’t a recommendation so far…) So why this cup? This cup is just good. I don’t know. It’s easy to assemble and clean (wide top). Doesn’t spill. Kids like straws?\n\n\n\nBooks\nNot that you’ll have time to read…\n\nThe Happiest Baby on the Block; Fully Revised and Updated Second Edition: The New Way to Calm Crying and Help Your Newborn Baby Sleep Longer. I’m sure there’s a quick YouTube video showing the 5 ways to sooth a baby. (Here’s an article.) But if you like books this is a good one. TLDR: sucking, swaddle, shushing, swinging, and side are all soothing to a crying baby.\n1-2-3 Magic: Gentle 3-Step Child & Toddler Discipline for Calm, Effective, and Happy Parenting. This is for toddlers, but I’ve learned this technique has been very helpful for giving me sanity in the terrible two phase. The approach doesn’t always work, but I’ve learned when it does, which is important. TLDR: don’t talk (engage in debate) with your toddler or show emotion when they’re misbehaving. Count them when they do bad acts. That’s 1. That’s 2. That’s 3. Break time. The book explains the method. YouTube is probably a friend here too.\n\n\n\nFinal word\nFatherhood has pushed my limits in all the ways. But I thank God every day I’m a dad. I do a nightly gratitude journal of my favorite moments of the day. If I’m paying attention during the day, at night it’s always my kids that have brought me the happiest moments and greatest joys of the day.\nJust remember: the clothes/products/trips are like the tools used to construct a fine painting. At the end of the day, it doesn’t matter what tools you use. Nobody cares and neither will you. It’s the output that matters most. I’ve found by just really paying attention to and getting curious about my kids in any moment helps me see the beauty of it all, while distractions cause me to think I need this or that be a better dad.\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Hamilton is running out of time\n\n\nI’ve been writing for my whole life. I have at least 15-20 journals. I have 3000 notes in one app, 2200 notes in another, at least 1800 digital journal entries over 10 years, I’ve built and replaced probably 5 blogs. This blog has 57 entries so far. What have I learned?\n\nMake writing daily as easy as possible. I set up some automation so I can write from my phone for daily blogging.\nWriting makes things feel real for me. It helps connect the dots. It’s a far more productive way to read the news. Less stuff, more thinking. Because if I can’t write at least one useful note about something I read, I realize it wasn’t worth reading.\nPerhaps reading is the more important daily duty. Well, to write I need something to inspire me. Usually that comes from the news. Sometimes inspiration comes from yesterday’s post. This inspiration came from editing my prior post.\n\nI’m not gonna worry about who reads this. But instead focus on writing every day because…\n\nI think it helps me practice storytelling\nhelps me practice communication\n\nHamilton wrote like he was “running out of time”. I’m writing because I want to use my time how I enjoy it.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2020-12-16-pretotyping-ml-solutions/index.html",
- "href": "posts/2020-12-16-pretotyping-ml-solutions/index.html",
- "title": "How to mock the value of an ML solution",
+ "objectID": "posts/2023-12-21-the-dodo/index.html",
+ "href": "posts/2023-12-21-the-dodo/index.html",
+ "title": "The Allegory of the Dodo",
"section": "",
- "text": "IBM once launched a demo of speech-to-text in the early day’s of the technology. But there was a twist: it was smoke and mirrors. A typist in behind the stage was translating and typing.\nThis is a mechanical Turk.\nIn data science, we can present results or dashboards to audience members before they see anything. If you see a big reaction, you know you found a number they cared about.\nGo ahead, write the conclusion first. Fake the chart first. Then back out the code that could produce the insights they need.\nSometimes, unique insights appear along the way. But most of the time you can properly gut-check whether someone will care about what you’re doing. Save time. Work backward.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "The scoop\nThere’s a great kids book, called “There’s a dodo on the wedding cake”.\nSo there’s this dodo, right, who walks in and hops on a cake. The attending violinist then scares it away and is praised for protecting the cake.\nTo the reader’s delight, more animals start appearing and the violinist excitedly protects the cake through hilarious misadventure. (My favorite is when he says, “begone beavers!”)\nTwo funny things happen though:\n\nThe guy soon realizes that all these creatures are wedding guests.\nThe guy has such a big nose and he’s allergic to marshmallows, that when he’s offered some cake he sneezes and the cake goes flying.\n\n\n\n\ndodo on a cake\n\n\n\n\nMorale?\n\nWhen you praise someone for doing something good, they do it again and again. → There’s a natural hero inside all of us.\nBut only the first animal was a problem. The beavers were the chefs. So he ends up creating all these problems thinking they’re problems. → In tech, when you fix a bug you then think other things are bugs and go off fixing those too. No, just do the first. Then check in.\nThe irony is that, despite saving the cake, he sneezes and ruins the cake. → Despite all the effort, the greatest protection to the cake was him staying away from it. Sometimes by trying to solve a problem we can make things way worse and create all these problems. (Politics? Create problems by creating solutions to problems that need solutions.)\n\nDon’t mistake a beaver for a dodo. But don’t mistake a dodo for a wedding guest.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2023-12-19-how-to-encourage-your-kids-/index.html",
- "href": "posts/2023-12-19-how-to-encourage-your-kids-/index.html",
- "title": "How to encourage your kids",
+ "objectID": "posts/quarto-from-phone/index.html",
+ "href": "posts/quarto-from-phone/index.html",
+ "title": "How to blog from your phone using quarto (or hugo/jekyll/etc) using Working Copy and Obsidian",
"section": "",
- "text": "Foster talent by:\n\ncreating stress free environment\nfinish before they’re tired\ndon’t pamper them at the country club\ntreat them for 10 minutes after\nno need to teach. Kids will figure it out\npeople who are great don’t necessarily love what they do, but those who love what they do will be great\n\n\n\nThis from Padraig Harrington should be required viewing for parents. Some really great advice.pic.twitter.com/ExE5rZpg7a\n\n— Kendall Baker (@kendallbaker) December 18, 2023\n\n\nI love it.\nThis is probably good management advice in any capacity.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Here are the steps I use to post content from my phone to bryanwhiting.com.\n\nCreate a new, empty vault in Obsidian\nOpen working copy, set up synced repo. This means any changes are auto-synced.\nAdd remote, point it at your GitHub repo where your blog is\nSet up a “pull, stage, commit, push” shortcut\nSet up GitHub actions to build, and push (use my repo’s workflow.yaml if you need an example)\nHave GitHub pages (if your repo is private) or Netlify (if your repo is private, because you can’t have a private GitHub repo without paying)\n\nAnd that’s it! Write markdown in obsidian. Push when you’re ready, let GitHub actions build and deploy. Check GitHub actions logs for any errors! (You’ll get an email if it fails.)\n\n\n\nSet up synced repo\n\n\n\n\n\nLink the repo\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2022-07-03-when-data-collection-goes-too-far/index.html",
- "href": "posts/2022-07-03-when-data-collection-goes-too-far/index.html",
- "title": "When data collection goes too far",
+ "objectID": "posts/2023-12-12-future-of-coding/index.html",
+ "href": "posts/2023-12-12-future-of-coding/index.html",
+ "title": "Future of Coding",
"section": "",
- "text": "Self Surveillance\nI came across this article in Flowing Data and thought it was fascinating.\nI’ve been on a productivity kick lately and notice these apps have habit trackers. I’ve tried using them in the past and have realized it gets too much too quickly. But that doesn’t mean as a data guy that I don’t think it’s a cool idea. It just overwhelms me.\nWe’ll I discovered that me tracking a few habits a day is junior varsity level. Felix’s blog is how it’s done to the extreme. Clearly this guy isn’t worried about Big Tech or Big Government harvesting his personal data. I love the boldness!\nThis is just 1/20th of the full page screen shot:\n\nThe number of slices and views he has is endless. It’s crazy to see everything he’s tracking, partly I’m sure because he has fun with it and the other part because he deep down hopes there’s something there.\nHis quote on Flowing Data has stuck with me:\n\nOverall, having spent a significant amount of time building this project, scaling it up to the size it’s at now, as well as analysing the data, the main conclusion is that it is not worth building your own solution, and investing this much time. When I first started building this project 3 years ago, I expected to learn way more surprising and interesting facts. There were some, and it’s super interesting to look through those graphs, however retrospectively, it did not justify the hundreds of hours I invested in this project.\n\nThis guy probably knows himself in a way that none of us ever will. He can’t ever lie to himself and say “yea, I’m good at exercising every day” because he’s got the data to back it up. But is this the life we want?\n\n\nSelf Surveillance as a Service\nSSaaS? Apparently. Exist.io tries to find patterns in your personal data and tell you when your most happy.\n\n\n\nExist.io\n\n\nWhen I researched the Apple Watch team a few years back I came to appreciate what they were/are trying to do: correlate heart monitoring with insurance claims data to proactively identify preemptive signs of cardiac arrest and other health issues.\nThat seems meaningful in theory because information about other people can be correlated with information about me.\nBut does info about myself correlate with info about myself? Probably. I have cycles. I have common traps and pitfalls like any human walking this earth. Some pitfalls are daily, some are quarterly or seasonal.\nBut I can also take 5 minutes to ponder reflectively or talk to a close friend and identify those patterns. What’s more human?\nAnd what’s more: if you end up changing your behavior then your past can’t really predict your future. So all of the data becomes moot once it achieves its purpose: true lasting change. Because if you truly change your behavior based on the insights you glean about yourself, those past triggers and anomalies no longer affect you in the way they did. So when you experience a personal step change in life, you’re a new you. Maybe not. But maybe.\nAlso, the new generations growing up on smartphones have a unique challenge: their online identities are given to them before they’re really conscious enough to determine if that’s what they want to show the world. Parents posting YouTube or TikToks of their kids are cementing for these kids an identity they themselves didn’t choose. We’ve all gone through phases of deleting Facebook photos from 15 years ago because it doesn’t reflect our current self image. What’s to be said of giving people an opportunity to change if everything they’ve ever said or done is recorded somewhere? Twitter. Instagram. TikTok. Facebook. Every post is cementing us in a way.\nI think it’s better to forget yesterday. Because it doesn’t have to have any bearing on where you’re going tomorrow. It can if you let it. But your past doesn’t have to define you.\n\n\nTakeaways\nWould you rather spend hours setting up digital automatic and manual tracking systems just to learn what you could by opening a gratitude journal every day? Sometimes less is more. I don’t think digital self surveillance is a direction we should go. (Not to mention the moment you do get hacked you’ll deeply regret it.)\nEven blogging can be dangerous for your career. Never know what’s gonna come back to haunt you years later.\nSo I should be careful what I say here, that is if I ever care to work for a self surveillance company in the future.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "The Essence\n\n\n\n\n\n\nTL;DR: The future belongs to those whom it’s always belonged\n\n\n\n\n\n\nA blogger data scientist Santiago recently said on Twitter, (paraphrasing) “I’m going to focus on what won’t change”. He was worried about how coding and AI will be automated, but Jeff Bezos wasn’t worried about so much of the world changing with the Internet. Bezos focused on what wasn’t changing.\nThis thought hit me, because I think it’s relevant to what our future looks like in 18 months.\n\n\nEvery day, a new winner\nNothings more apparent that the world is going to be wildly different than the past than to hop on X and see new academics one upping themselves with the next greatest LLM. Mistral just announced a 2Bn valuation as a 6 month old company. Fast is the new pace. And it’s oddly the case that big businesses are capable of massive disruption from upstarts. They can’t afford to stay glacial much longer (5-10 years).\nWith all of this, what about this blog post won’t change? Well, the timeless principle is that the nature of the game has changed permanently. Pandora’s box was opened.\n\n\nThe rise of a new equalizer\nGo into a poor part of town and you’re guaranteed to see one thing: poor people using smart phones.\nThey have access to all the same information that the rich people have. The same excellent software. Same excellent hardware. All for a fee of $50/mo for the hardware and $25/mo for an unlimited data plan.\nSo that out the rich and the poor on the same level.\nWell now, knowledge and reasoning is as accessible as Google Maps. You don’t have to pay $20/mo for ChatGPT plus. You can use Bing, Claude.ai, bard.google.com, You.com, or perplexity.ai for free. Now.\nSure people hate on bard and want to use the latest and greatest. But even the crappiest of the four I mentioned above (granted Bing/You.com use GPT in the background), is still leaps ahead of where we were just a year ago. If we were stuck only using one of these tools forever we’d still be massively more productive than not.\nSo knowledge and intellectual reasoning is going to be commonplace amongst the plebes.\n\n\nif you value intelligence above all other human qualities, you’re gonna have a bad time\n\n— Ilya Sutskever (@ilyasut) October 7, 2023\n\n\n\n\nMore of the same\nSo…what’s going to happen? Well, I’m not sure human nature will change much.\nThey say “being rich doesn’t make you happier. It just makes you more of what you already were before being rich”.\nAnd so, the mentally poor will remain so. They’ll use AI for entertainment. Better video games. Better, faster content. More consumption.\nThe creatives will be more creative. Nobody will be at a barrier of being silo’d. So what if you’ve only done data science for 10 years. You know coding. It’s easy to now do web design. Or copy writing. Or art. Or…\nAnd I won’t be able to stop the masses from using Code Interpreter to do data science. The systems will only become more powerful, so cloud companies will make it easier to write sql or whatever. Heck, 5-10 years English will be the only programming language. (Who here does firmware/bit mapping anymore? Only a select few.)\n\n\nThe hottest new programming language is English\n\n— Andrej Karpathy (@karpathy) January 24, 2023\n\n\nPython programmers will become like the engineers who create the seeds. And the rest of us just get bigger, better, and more efficient farms. There will always be those who specialize, as there are people who write compilers today, but fewer and farther between. Perhaps intellectual interest and economic incentives will equalize.\n\n\nFestivus for the rest of us\nCompanies will become smaller. But the same human needs will need to be met:\n\nfood\nshelter\nclothes\nphysical toys\ndigital entertainment\n\nMore people will just be able to do more things. Perhaps there will be more prosperity as a result.\nAnd more people will unionize and try to prevent the change. But that’s a short term thing. Can’t last forever. A union is only as good as it’s needed.\n\n\nTakeaways\nGet ahead by focusing on what’s essential. Leverage what’s around you to develop new skills to prepare for the future. And be excited. It’s going to be a wild ride.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2023-12-25-ai-robots-are-mastering-tasks-faster-than-humans-/index.html",
- "href": "posts/2023-12-25-ai-robots-are-mastering-tasks-faster-than-humans-/index.html",
- "title": "AI Robots are mastering tasks faster than humans",
+ "objectID": "posts/2020-12-18-productivity-notes/index.html",
+ "href": "posts/2020-12-18-productivity-notes/index.html",
+ "title": "Productivity notes",
"section": "",
- "text": "With a proper objective function, machines can do really hard, isolated tasks\n\n\n\n\nthey can walk\nthey can solve complex physical challenges\nthey can play the toughest strategic games on earth and win (AlphaGo 2017, AlphaStar 2019)\nbut can they determine the objective function by themselves?\n\n\n\n\nReinforcement Learning\nGreat article written at Techradar going over the advances in reinforcement learning that allow a robot to finish labyrinth faster than humans. The video does the feat even more justice:\n\nThis stands out to me in a few ways, principally that they crippled the machine to not take shortcuts. It still won.\nReminds me first of AlphaGo:\n\nWhich reminds me of AlphaStar. Here’s a video with commentary on AlphaStar’s performance:\n\nThe thing I connect this video (or some other I watched) with labyrinth is that AlphaStar devised brand new strategies. Players against AlphaStar remarked how they felt like it was indeed being creative. Likewise, Cyberrunner (labyrinth) found shortcuts.\nAll of this points to reinforcement learning as a viable solution to getting stuff done both in the digital and physical worlds. Take, for example, how RL is used to teach a robot how to walk.\n\n\n\nAn AI robot learning to play StarCraft for the first time…\n\n\n\n\nWhat’s the objective?\nThe goal for Super AI will be to figure out what the objective is. If we can marry digital and physical worlds, teaching the systems how to solve objective functions, the possibilities become…super/human.\nThe unique thing about a human, it seems, is that we have our objective function hardwired into us: survive. First this is “food/shelter”. Next it’s social survival.\nHow do humans figure out objective functions so fast? A human looking at labyrinth will know what to do within a second. Does it have interest/patience to sit there and learn it? Does GPT 4Vision have the ability to figure out these objective functions?\nWe learn by objective functions. Minimize wrong answers on a test. Minimize missed baskets on the basketball court. Minimize rejections in a sales funnel. Maximize great conversation and connection at a dinner party.\nHumans learn very quickly. For now, robots are being isolated to specific tasks and their results are impressive. But perhaps to become our overlords all they need is the one objective function to rule them all. Er, uhm, and by “them” I of course don’t mean “us”…\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Time Management\n\nCalendar my ToDos.\nShorter work is better. Less to edit. Less mistakes. Keep it simple.\n\n\n\nDistractions\n\nBlock 30 min for email and review team members work twice a day.\nAggressively filter email spam.\n\n\n\nAnalysis\n\nStart with a story. End with story.\nIterate quickly with stakeholders.\nalways have one nights sleep before sending a presentation. Edit.\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2023-12-13-learn-by-doing/index.html",
- "href": "posts/2023-12-13-learn-by-doing/index.html",
- "title": "Learn by doing",
+ "objectID": "posts/2018-07-02-semantic-versioning-for-data-science-models/semantic-versioning-for-data-science-models.html",
+ "href": "posts/2018-07-02-semantic-versioning-for-data-science-models/semantic-versioning-for-data-science-models.html",
+ "title": "Semantic Versioning for Data Science Models",
"section": "",
- "text": "TL;DR: Yes, unintelligent people can use things without fully understanding them\n\n\n\n\n\n\n\n\nMany people have their noses up their ass when they claim you can't use an algorithm you don't fully understand.They give themselves this fake moral superiority. They must be the only ones who belong because they know every detail.This is all bullshit.Here are a few simple…\n\n— Santiago (@svpino) December 12, 2023\n\n\nWhile the language is perhaps a bit strong, I agree with the sentiment here.\nI don’t have a PhD but have been limited twice in my career by those who wish I had one. What would having a PhD give you that I don’t have?\nWell, I trust you more if you had one.\nJeremy Howard, a world renowned research and creator of fast.ai and answers.ai started his lectures on deep learning with “you don’t need to have a math degree, you just need to know a little bit of code”. With that philosophy he’s helped entrepreneurs create amazing products that left his free online program.\n\nBlindspots tho\nConcepts are important to understand. Understanding your training data and how the machines learn from them and the biases they’re prone to is very useful, perhaps mission critical, to creating value.\nSo what’s the line? Well, if you got a PhD you’d probably spend 80% of your time on your dissertation and taking some non relevant coursework. How does that help you today?\nWell, you should learn as many concepts as you can. Like driving a car, you don’t need to know hydroplaning but you should learn hydroplaning. You don’t need to know how to drive in snow but you should know that driving a car in the snow with rear wheel drive is a different experience.\nIf you live in Arizona you don’t need to know how to drive in the snow. Your bias is you’ll never know why that would be necessary.\n\n\nPrinciple\nBe so curious that you explore the boundaries of your space such that you learn “oh, cars work differently in snow, but I don’t have snow so I’m safe”. Once you reach the boundary you probably know enough. Then ship your code and learn by how you messed things up.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "If you’ve ever wanted to tag your data science model, you’ve probably wondered how to version it. Which will it be: vx.4.1, v34.1231.51.21, or v91.x4.dev34? After reading about semantic versioning, I propose a method for versioning data science models.\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html",
- "href": "posts/2022-06-23-the-best-todolist-manager/index.html",
- "title": "What is the best todo list manager?",
+ "objectID": "posts/2018-07-02-semantic-versioning-for-data-science-models/semantic-versioning-for-data-science-models.html#an-example",
+ "href": "posts/2018-07-02-semantic-versioning-for-data-science-models/semantic-versioning-for-data-science-models.html#an-example",
+ "title": "Semantic Versioning for Data Science Models",
+ "section": "An Example",
+ "text": "An Example\nI build data science models by building python packages and committing the code to GitHub. The python package contains all of the support files and a main.py file to run the pipeline from start to finish. The support files could contain either helpful loggers to tell me where the pipeline is breaking down, diagnostic tools such as an AUC-ROC plot, files to engineer features, or files to train different model types. So there are really these three things that could happen to my code at any time. It doesn’t matter where those changes occur, or how many lines of code changed. What matters is what’s happened to the model. Each time I get a pull request approved, I’ll update the version number in my repo.\n\nChanging BETWEEN\nIncrement the BETWEEN version for the following changes:\n\nWhen the structure of your data changes\nIf your target variable changes (how you coded it, or data that produced the target changed)\nThe underlying population you’re training on changes\n\nAssume I discovered that I had a bad join. The join change might have been a minor one, but it’s now difficult to really compare the two models because my target was affected as a result. What I was predicting, even if slightly different, is now something different. If you ever feel like you’re comparing apples to oranges when looking at an AUC-ROC curve, update the BETWEEN version.\nWith each BETWEEN change, there should be a clear communication in the release notes of why the model is inherently different from prior models. These changes should be less frequent.\n\n\nChanging WITHIN\nIncrement the WITHIN number when the following happens:\n\nFeatures are added\nData sources are added or updated. (You might have several data sources today that help contribute to the target. But adding a data source might just mean adding new features. If you add data in such a way that it changes what you’re ultimately modeling, update the BETWEEN version.)\nNew modeling types are added. (You might be using a logistic regression, but add a mo)\n\nSo if I added files that allowed me to train different model types, or added files that engineered new features, I’d incrememnt the WITHIN version. In my mind, this shouldn’t be a count of features you include, but should just be incremented every time you do something that affects the model performance.\nAt any time, you might what to view different subsets of features and their effect on the model. So, you might not change the number of features, but add some functionality into your pipeline that now produces three different models in each run. I would increment the WITHIN version number in that case.\nRemember, the goal of this type of versioning is to show generally what’s changed in your pipeline.\n\n\nChanging PROCESS\nIncrement the PROCESS version when the following happens:\n\nA minor bug is fixed (unless this updates BETWEEN or WITHIN)\nEnhance a logger or aspect of the pipeline\nAdd a diagnostic plot or table\n\nBasically, any time you add something that doesn’t ultimately affect the performance of the model, you should update the PROCESS version. Your changes might improve the flow of your code and make you a much happier developer, but if it doesn’t improve the model, then the key priority of your code hasn’t improved."
+ },
+ {
+ "objectID": "posts/2023-12-15-prompt-techniques/index.html",
+ "href": "posts/2023-12-15-prompt-techniques/index.html",
+ "title": "Prompt techniques",
"section": "",
- "text": "Wirecutter reviewed the best todo list apps and declared “ticktick” the winner. After all my research here’s what I’ve learned.\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "How to Prompt Engineer\n\n\n\n\n\n\nTL;DR: Prompt Engineering\n\n\n\n\n\n\n\n\nOpen AI released a great Prompt Engineering guide (link in comments). Here is a summary of their 6 strategies for getting better results when prompting GPT-4 pic.twitter.com/QiBASsbOHD\n\n— MindBranches (@MindBranches) December 15, 2023\n\n\nThe summary above is from here:\nPrompt Engineering\n\n\nTakeaways\nIf prompt engineering is the future of coding, then coding is going to change from a deterministic, imperative experience to a stochastic one. That seems odd. 95% chance your output code will do what you want it to do, with a 5% change it does something drastically wrong. How do you unit test that?\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#structured",
- "href": "posts/2022-06-23-the-best-todolist-manager/index.html#structured",
- "title": "What is the best todo list manager?",
- "section": "Structured",
- "text": "Structured\nThe highest rated app for planning on the App Store is “Structured”. Where this app excels is it’s simplicity. It’s likely best for someone like a student that wants to plan out when they’ll get their homework done. What I loved about this app, and what I wish many others apps had, is a duration estimation for a task. Why is it so hard for developers to realize that tasks take time? They obsess over list keeping and app design but ignore the second most basic question: how long will this take? Structured kind of solves this. What I don’t like about this app is the design. While others praise it for its beauty and simplicity I thought the visual feel of it was distracting."
+ "objectID": "posts/2022-11-12-what-is-a-model/index.html",
+ "href": "posts/2022-11-12-what-is-a-model/index.html",
+ "title": "What is a model?",
+ "section": "",
+ "text": "_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#microsoft-to-do",
- "href": "posts/2022-06-23-the-best-todolist-manager/index.html#microsoft-to-do",
- "title": "What is the best todo list manager?",
- "section": "Microsoft To Do",
- "text": "Microsoft To Do\nMicrosoft To Do used to be a favorite of mine. I loved its daily planner and smart suggestions. It has the philosophy that every day you should start fresh. All unfinished tasks will be removed from today and you’re left to pick and choose what you want to do. The fatal limitation however is that it lacks the ability to plan tomorrows a tasks the day before. So this forces people to have to populate their tasks at the start of the day. This is the worst time for planning for me because I want to get up and get going ASAP. (But honestly most times I wake up 15 minutes before my first meeting or am spending time pouring milk into my kids’ cereal.) Morning is not the best time to plan a day for me, so I lost interest here.\n\n\n\nTo Do has smart text parsing.\n\n\n\n\n\nTo Do has “smart suggestions” that you can add to your day."
+ "objectID": "posts/2022-11-12-what-is-a-model/index.html#im-building-a-model-you-know",
+ "href": "posts/2022-11-12-what-is-a-model/index.html#im-building-a-model-you-know",
+ "title": "What is a model?",
+ "section": "I’m building a model, you know",
+ "text": "I’m building a model, you know\nThe first time I heard the word “model” was from an engineering professor in college. I had no idea what he meant so I asked him. He looked at me like many who have forgotten their own ignorance and just said “Some calculations in excel. You know.”\nNo, you don’t know. But nod your head so you don’t look like an idiot. “Right, of course.”\nWhy this moment was intriguing to me is one of my personal mysteries. But it was one of those odd moments that stuck. I’m now here to help my old self make sense of the world."
},
{
- "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#todoist",
- "href": "posts/2022-06-23-the-best-todolist-manager/index.html#todoist",
- "title": "What is the best todo list manager?",
- "section": "Todoist",
- "text": "Todoist\nI’ve tried many times to get into Todoist. It has smart text parsing which is apparently best in class from my Reddit research.\nThings it has:\n\nProjects can have section labels. This lets you organize your list more meaningfully. This is great, only that TickTick does it better. Each project in TickTick can have subprojects. These subprojects act like sections. You can also move subproject from one list to another.\n\n\nThings it lacks:\n\nCan’t nest labels (tags)\nCan’t create labels dynamically - you need to manually add them on the label screen before a task can leverage them.\nCan’t pull in my calendar.\nDoesn’t have check items, only subtasks."
+ "objectID": "posts/2022-11-12-what-is-a-model/index.html#the-same-name-for-different-things",
+ "href": "posts/2022-11-12-what-is-a-model/index.html#the-same-name-for-different-things",
+ "title": "What is a model?",
+ "section": "The Same Name for Different Things",
+ "text": "The Same Name for Different Things\nWhen I told my mom I build models for a living she shook her head. Turns out my dad had been telling her for years that he builds financial models, yet she never figured out exactly what it was that he did.\nLet’s be clear on the common ground. A “model” per the dictionary is “simplified description, especially a mathematical one, of a system or process, to assist calculations and predictions.” By this definition, my dad and I do the same thing. We use numbers to put structure on the world."
},
{
- "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#sorted3",
- "href": "posts/2022-06-23-the-best-todolist-manager/index.html#sorted3",
- "title": "What is the best todo list manager?",
- "section": "Sorted3",
- "text": "Sorted3\nThis app takes a different design. It’s main selling point is the ability to auto schedule your day based on the tasks you have for that day.\nPros:\n\nRich formatting on description. Supports markdown.\nclean UI. Good shortcuts.\nintegrates with calendar to plan your day.\nEasily plan duration.\n\nReally nice widgets.\n\nCons:\n\nDoesn’t have subtasks, but has check boxes\nNo prioritization or smart filters. Just tags and lists.\nCan’t easily see across lists like you can in TickTick\n\nAuto schedule is nice but could be kind of gimmicky. If you have 8 meetings in a day and five 30 min free blocks but all your tasks for the day are an hour, it would put all your tasks at the end of the day. I found it kind of buggy."
+ "objectID": "posts/2022-11-12-what-is-a-model/index.html#types-of-models",
+ "href": "posts/2022-11-12-what-is-a-model/index.html#types-of-models",
+ "title": "What is a model?",
+ "section": "Types of Models",
+ "text": "Types of Models\nA financial model is different than a statistical model is different than a physical model or a 3D architecture model. But they all do the same thing: use numbers to put structure on the world.\nEvery model has a few things in common: - They make assumptions about the world - They use data - They probably make some calculations to help someone make a decision\nA financial model might make assumptions in a spreadsheet about how many customers you’ll get per month for the next 12 months. This doesn’t seem complicated, and it shouldn’t: models don’t have to be complicated.\nA data science model is one that might try to predict whether a customer is going to click on an ad."
},
{
- "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#automated",
- "href": "posts/2022-06-23-the-best-todolist-manager/index.html#automated",
- "title": "What is the best todo list manager?",
- "section": "Automated",
- "text": "Automated\nMotion became popular as a calendar organizer. Add a task and Motion will figure out when to do that for you. If you change your meeting then Motion will adjust when your tasks will get done. Cons: it’s expensive ($20-$35 per month). It’s mobile app is horrid. It didn’t seem like it offered sub tasks and all that. So you really need to trust the algorithm.\n\nIt gets the notion of task-duration right. But if you want to jot down notes of all the next actions you need to take on a project, your up the creek. I’m also not sure how it handles task dependency - if I put in 5 tasks for a project will it know their order? Not sure. Again I’ve never tried this app so take this all with a grain of salt.\nReclaim is another app claiming to do the same thing and offers a free tier. I did as much as set up my account with this.\nPerhaps if I gave these more attention they’d prove their worth. I’ve tried calendaring my todos in the past and it hasn’t been very effective for me. Perhaps these apps would solve it but I think I need a good mobile app.\nAnyway, you can’t beat free and free is working for me on TickTick right now. Also, my calendar is very open generally so I don’t have the same problems a busy manager might have.\nI think these apps would be best for people who have meeting heavy workflows. It handles the schedules as you need."
+ "objectID": "posts/2022-11-12-what-is-a-model/index.html#how-does-a-model-work",
+ "href": "posts/2022-11-12-what-is-a-model/index.html#how-does-a-model-work",
+ "title": "What is a model?",
+ "section": "How does a model work?",
+ "text": "How does a model work?\nThe purpose of a math model is to imitate the “data-generating process”. Meaning, if you have some data on your customers (what they purchased and when), you would hope that there’s some rules that your customers are operating under. If you knew their income, their budget, their preferences, etc then you can guess how much they have to spend at a given time. Thus, if you know how much gas is in their tank, the mileage of the vehicle, the MPG of the vehicle, you can predict when they’ll stop to fill up on a road tip. When they fill up, they generate some data. “Bob purchased $34.52 of gas on Monday at 5:43pm.” Why did they purchase at that time of day? And why did they spend exactly $34.52?\nIf all you have outcome/results data then you can’t ever make predictions. But if you know what type of car Bob drives, how often he drives, how much gas he had in the morning when he left his house, etc, you can start to predict when he’ll next fill up.\nThe assumption behind a model is that what was true in the past will remain true in the future.\nWhy have there been so many issues because of COVID? Nobody could have predicted a world wide pandemic. The world wasn’t the same. All the models stopped working. Supply chain models. Financial models. Default models. All of these assumptions across all industries started breaking down."
},
{
- "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#akiflow",
- "href": "posts/2022-06-23-the-best-todolist-manager/index.html#akiflow",
- "title": "What is the best todo list manager?",
- "section": "Akiflow",
- "text": "Akiflow\nThis app seemed expensive ($15/mo), no mobile app, but has an awesome calendar. The philosophy is you have to assign every task to a date and time to move it out of the inbox. That seemed like a headache to me, and seemed like I’d be constantly managing the inbox. That and I’m cheap and want a mobile app."
+ "objectID": "posts/2022-11-12-what-is-a-model/index.html#how-do-you-build-a-data-science-model",
+ "href": "posts/2022-11-12-what-is-a-model/index.html#how-do-you-build-a-data-science-model",
+ "title": "What is a model?",
+ "section": "How do you build a data science model?",
+ "text": "How do you build a data science model?\nMost people build models every time they calculate the “mean”. This is the most ubiquitous model and anybody can use it. Sum your revenue and divide by number of purchases.\nIf you want to predict revenue for next Monday, just take the average revenue for every past Monday. That’s a model.\nIf Monday is a holiday or the Monday after Black Friday, your model might be very wrong. So you estimate black Friday’s sales by looking at last year’s sales. “Last Cyber Monday we sold 2x more than we typically do on a Monday. So I’ll predict that on this upcoming Cyber Monday I’ll do 2x more than what I did in the most recent Monday”\nNow that you’re taking in some data, your building a model. You’re translating a theory on how the world works into numbers. You’re taking last year’s sales and using it to predict tomorrow’s sales."
},
{
- "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#sunsama",
- "href": "posts/2022-06-23-the-best-todolist-manager/index.html#sunsama",
- "title": "What is the best todo list manager?",
- "section": "Sunsama",
- "text": "Sunsama\n$20 per mo or $192/year. Super fast customer support.\nFeatures: * Mobile app, Mac app. * Duration\nPros: * Channels (lists) * great shortcuts * can align tasks with objectives * more of a mindful daily planner * Integrates with Todoist and other apps. You can pull in all to foist metadata including subtasks and check them off in Sunsama. Additionally, you can add duration and calendar your Todoist tasks. * Can export all the tasks to csv. * Can split calendar events.\nCons: * Widgets are weak * No priority labels * No task tags/labels. (But you can use Todoist for this if you want.)"
+ "objectID": "posts/2022-11-12-what-is-a-model/index.html#start-simple",
+ "href": "posts/2022-11-12-what-is-a-model/index.html#start-simple",
+ "title": "What is a model?",
+ "section": "Start Simple",
+ "text": "Start Simple\nWhenever I build models, I always try to start simple and see if something more complicated can beat it.\n\nUse averages\nBuild a rules-based model\nBuild a regression model/machine learning model\nBuild a ML model with rules on top of it\n\nIf I’m trying to predict whether a user will click an ad, the simplest model is “what % of all users clicked this ad in the past?” If 20% of all users who see an ad click on it, then my “model” is “predict a 20% chance of click”.\nThen to make it more complicated, you can segment your users. “10% of men click on this but 25% of women do.” You now have more data and better predictions.\nA “decision tree” is a machine learning algorithm that just finds those segments in your data. “If man age 30 in Chicago then 22.3% chance of clicking based on all past behavior of all 30-year old men in Chicago”. It’s called a decision tree because it will put every user into mutually exclusive groups.\nSome “business logic” model might be making predictions around how your business operates. For example, if there’s a sale you may know that clicks increase."
},
{
- "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#reclaim.ai",
- "href": "posts/2022-06-23-the-best-todolist-manager/index.html#reclaim.ai",
- "title": "What is the best todo list manager?",
- "section": "Reclaim.ai",
- "text": "Reclaim.ai\nTask manager isn’t as good as Sunsama. Only $8/mo for pro. Has todoist integration too, but that’s a pro feature.\nThis is one I’ll want to try later. It has the habit tracking and smart flexible calendaring. Not sure how good of a task manager it is. TBD."
+ "objectID": "posts/2022-12-15-chatgpt-rshiny/index.html",
+ "href": "posts/2022-12-15-chatgpt-rshiny/index.html",
+ "title": "Building a complex R Shiny Dashboard Using ChatGPT",
+ "section": "",
+ "text": "ChatGPT is nothing short of amazing, but I think I’ll still have a job as a data scientist for at least another year or so.\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#vimcal",
- "href": "posts/2022-06-23-the-best-todolist-manager/index.html#vimcal",
- "title": "What is the best todo list manager?",
- "section": "Vimcal",
- "text": "Vimcal\nFast calendar. Not for tasks or backlogs it seems though. Didn’t try."
+ "objectID": "posts/2022-12-15-chatgpt-rshiny/index.html#data-processing",
+ "href": "posts/2022-12-15-chatgpt-rshiny/index.html#data-processing",
+ "title": "Building a complex R Shiny Dashboard Using ChatGPT",
+ "section": "Data Processing",
+ "text": "Data Processing\nI asked it to create an indicator variable for me so that I could color my dygraphs chart by forecasts vs. historical:\n\nHere was my prompt:\n\nDon’t use add_series() since you’re wrong and it’s not a function in dygraphs. Instead, append the forecasts to the aapl in the stock_subset function and create a new indicator for the forecasted values where 0 is past data and 1 is forecasted data. Then, color the dygraphs chart based on the indicator value. Re-write this script below accordingly.\n\nLook at that code! It knew to filter appl2$ds > max(appl$ds). That’s exactly how you subset a prophet dataframe (which returns all historical and future forecasts in one dataframe). It knew that appl was the first dataset and appl2 was the derived dataset. It knew that it had to take the max date from appl. This is absolutely amazing, hands down.\naapl2 <- bind_rows(aapl2, forecast_subset)\naapl2$indicator <- ifelse(aapl2$ds > max(aapl$ds), 1, 0)"
},
{
- "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#asana",
- "href": "posts/2022-06-23-the-best-todolist-manager/index.html#asana",
- "title": "What is the best todo list manager?",
- "section": "Asana",
- "text": "Asana\nOh man, how many times have I downloaded Asana? I started with it in 2014 and thought it was cool. Trying it again in 2022 it had a similar feel.\nAsana is the only task manager app I reviewed that has dependency management. Sure, Jira has blockers and all that but Jiras more aimed toward high level project management. It’s not fast at the\nOther things I remember liking:\n\ninfinite subtasks. This was fun and also distracting.\nAutomation features. You can create workflows for what happens to a task\nCommunication. It has a nice inbox feature that let me communicate with my manager easily.\n\nCons:\n\nClunky on the board management. If I want to start a new project things get messy and overwhelming."
+ "objectID": "posts/2022-12-15-chatgpt-rshiny/index.html#data-inspection",
+ "href": "posts/2022-12-15-chatgpt-rshiny/index.html#data-inspection",
+ "title": "Building a complex R Shiny Dashboard Using ChatGPT",
+ "section": "Data inspection",
+ "text": "Data inspection\nI then got this error:\nI'm getting this error: Error in `select()`:\n! Can't subset columns that don't exist.\n✖ Column `date` doesn't exist.\nChatGPT gave me the ring-around. It didn’t know that the dates were stored as the rownames, but confidently gave me answers like this:\n\nTo it’s credit, it recommended I inspect the data myself:\n\nOnce I figured out the problem, I told it that the rownames were dates, and that date wasn’t a column. It had just the solution for me:"
},
{
- "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#jiragithub-projects",
- "href": "posts/2022-06-23-the-best-todolist-manager/index.html#jiragithub-projects",
- "title": "What is the best todo list manager?",
- "section": "Jira/GitHub projects",
- "text": "Jira/GitHub projects\nGreat at a high level but messy to manage the gritty day to day. Doesn’t let you schedule or put due dates. So eventually I’d need some other tool to manage all the minutia of getting things done."
+ "objectID": "posts/2022-12-15-chatgpt-rshiny/index.html#misleading-functions",
+ "href": "posts/2022-12-15-chatgpt-rshiny/index.html#misleading-functions",
+ "title": "Building a complex R Shiny Dashboard Using ChatGPT",
+ "section": "Misleading Functions",
+ "text": "Misleading Functions\nTwice, it told me that dygraphs had functions that it didn’t actually have. It even wrote code for me. For example, it recommended this to me:\ndygraph(stock_subset(), main = \"AAPL Stock Price\") %>%\n add_series(data = forecast_subset, col = \"red\") %>%\n dyRangeSelector()\nAnd it doubled down:"
},
{
- "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#not-considered",
- "href": "posts/2022-06-23-the-best-todolist-manager/index.html#not-considered",
- "title": "What is the best todo list manager?",
- "section": "Not considered",
- "text": "Not considered\nI didn’t look at Things3 or OmniFocus, which are considered of the best tools. Things3 was too expensive and seemed like it had many of the same features as TickTick, except I personally didn’t like the UI."
+ "objectID": "posts/2022-12-15-chatgpt-rshiny/index.html#data-conversion",
+ "href": "posts/2022-12-15-chatgpt-rshiny/index.html#data-conversion",
+ "title": "Building a complex R Shiny Dashboard Using ChatGPT",
+ "section": "Data Conversion",
+ "text": "Data Conversion\nIt told me to use the xts package so that dygraphs could read the data in. This was accurate - dygraphs cannot read in tibbles.\n\nBut it was misleading me a little bit. Again, I had to look at the data, because the direct code it gave me was missing columns that it assumed it had."
},
{
- "objectID": "posts/2023-12-21-github-copilot-levels-up/index.html",
- "href": "posts/2023-12-21-github-copilot-levels-up/index.html",
- "title": "GitHub copilot levels up",
+ "objectID": "posts/2017-02-10-ggplot-plotly/ggplot-plotly.html",
+ "href": "posts/2017-02-10-ggplot-plotly/ggplot-plotly.html",
+ "title": "The DataViz battle: Plotly vs ggplot2",
"section": "",
- "text": "TL;DR:\n\n\n\n\n\n\n\n\ncan debug failed terminal errors\nshows git diff of revisions\ncan explore workspaces now with @workspace (still lags behind Cursor, which indexes everything in your code base)\nauto-commit messages (I’ll take this)\nauto-pr messages (sounds good to me)\n\nMore here:\nhttps://code.visualstudio.com/blogs/2023/11/13/vscode-copilot-smarter\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "R users fall in love with ggplot2, the growing standard for data visualization in R. The ability to quickly vizualize trends, and customize just about anything you’d want, make it a powerful tool. Yet this week, I made a discovery that may reduce how much I used ggplot2. Enter plot_ly().\nFor this post, I assume that you have a working knowledge of the dplyr (or magrittr) and ggplot2 packages. I caveat that this post is backed with only 4-5 hours using plotly(), so some statements here may not be fully vetted.\nPlotly and ggplot2 are inherently for different purposes. plotly allows you to quickly create beautiful, reactive D3 plots that are particularly powerful in websites and dashboards. You can hover your mouse over the plots and see the data values, zoom in and out of specific regions, and capture stills. Here’s a basic histogram:\n\nset.seed(1)\nlibrary(plotly)\nplot_ly(x = ~ rnorm(1000), type = \"histogram\")\n\n\n\n\n\nAfter a brief dabble this week in plotly, I realized quickly the many advantages that plotly has over ggplot2.\nSeveral initial impressions:\n\nPlotly handles multiple wide data columns. I always find it annoying that to color different series in ggplot2, your data had to be in long format. Granted, it takes one simple melt() command to get the data into wide format.\nPlotly also handles long format (see below).\nCustomizing the layout (plot borders, y axis) is easier.\nCustomizing the legend is easier (in ggplot2 I’ve wanted to remove just one series, which isn’t always easy).\nDocumentation is better in Plotly.\nPlotly syntax is very intuitive (learning how aes() in ggplot2 works is tricky at first)\nPlotly also works for Python, Matlab, and Excel, among other languages.\nIt’s very easy to add new series and customize them (one line, one scatter, and one bar, for example)\nYou can use other fonts (which is possible in ggplot2, but I’ve never gotten to work on my Windows machine)\nYou can toggle series on and off by clicking the series name in the legend\n\nBenefits of ggplot2 over plotly:\n\nFacet wrapping is very easy in ggplot2. (I think you have to do subplots in plotly.)\nggplot2 is probably quicker for exploratory analysis.\n\n\nConverting ggplot2 into plotly\nAn additional benefit of plotly is that you can convert your ggplot() graphs into a plotly object.\n\nlibrary(ggplot2)\np <- qplot(x = rnorm(1000), geom = \"histogram\")\np\n\n\n\n\n\n\n\n\nThen, invoking the ggplotly(p) command, we see the transformation:\n\nggplotly(p)\n\n\n\n\n\nA draw back of ggplotly() is that if you do refined customization (like putting your legend on the bottom of the graph), ggplotly() doesn’t seem to pick this up by default.\n\n# ggplot with legend on the bottom\np <- qplot(\n data = iris,\n x = Sepal.Width,\n y = Sepal.Length,\n geom = \"point\",\n color = Species\n) +\n theme(legend.position = \"bottom\")\np\n\n\n\n\n\n\n\n# Plotly doesn't pick up the legend change\nggplotly(p)\n\n\n\n\n\nBut since Plotly also saves to an object, you can use the %>% notation to pipe and add additional plotting commands. This is similar to the + operator in ggplot().\n\np <- qplot(\n data = iris,\n x = Sepal.Width,\n y = Sepal.Length,\n geom = \"point\",\n color = Species\n) +\n theme(legend.position = \"bottom\")\np2 <- ggplotly(p)\n# Use the plotly layout() command for legend customization\np2 %>% layout(legend = list(orientation = \"h\"))\n\n\n\n\n\nThe legend doesn’t do exactly what we want, but you can manipulate the legend location manually using x and y coordinates. The orientation = 'h' setting in the docs puts the legend on the bottom for default plot_ly() objects. Graphing the same series, we see the legend at the bottom:\n\nplot_ly(iris,\n x = ~Sepal.Width,\n y = ~Sepal.Length,\n type = \"scatter\",\n mode = \"markers\",\n color = ~Species\n) %>%\n layout(legend = list(orientation = \"h\"))\n\n\n\n\n\n(You notice the Plotly X-axis title can get cut off1, so let’s put that +1 to ggplot2.)\n1 This may depend on your screen.Plotly seems very intuitive relative to ggplot2 in doing layout customization. Things that took me many iterations on StackOverflow to figure out, like adding a black line on y = 0, are built in to Plotly.\n\np <- plot_ly(iris,\n x = ~Sepal.Width,\n y = ~Sepal.Length,\n type = \"scatter\",\n mode = \"markers\",\n color = ~Species\n)\n# Put legend on bottom, change the x-axis range, and turn on the x-axis line. \n# Also, make the zeroline visible, and turn it red.\np <- p %>% layout(\n legend = list(orientation = \"h\"),\n xaxis = list(\n zeroline = T, # Turns x = 0 on\n zerolinecolor = \"red\", # colors x = 0 red\n showline = T, # Shows xaxis border line\n range = c(-2, 7)\n )\n)\n# Or, save parameters into a list. Use new fonts (a huge plus)\nf1 <- list(\n family = \"Arial, sans-serif\",\n size = 18,\n color = \"lightgrey\"\n)\nyax <- list(\n title = \"Sepal length\",\n titlefont = f1\n)\np %>% layout(yaxis = yax)\n\n\n\n\n\nThings I’d like to further explore:\n\nYou can export static plotly images out to file. My hypothesis is that Plotly images take longer to generate than ggplot2. So if I’m mass producing 30,000 plots (which I had to do last month), which is the faster approach? I would assume ggplot2.\n\n\n\nPlotly in RShiny Dashboards\nThe goal in learning Plotly was for me was to eliminate the Excel-VBA dashboard I created using for my manager. Excel has (some) benefits over ggplot2 static charts: you can easily hover your mouse over a series to see the data value, and most industry users know how to manage an Excel axes. Grated, you can build in an RShiny widget to allow the user to control the axes, but Excel comes with that knowledge base built-in. ggvis allows for the powerful library of Google charts, but I think for a reactive dashboard, plotly is a great way to go2.\n2 Of course, this is relative to the chart you’re trying to make.So Plotly solved the Excel problem for me. Now my manager can click and zoom to the parts of the graph that are interesting, and hover the mouse to see the values. Just use renderPlotly() instead of renderPlot() in the server.R file, and plotlyOutput() instead of plotOutput() in the ui.R file.\nMore info here: RShiny and Plotly\n\n\nRShiny vs Plotly Dashboards\nBoth RShiny and Plotly allow for creating dashboards. Plotly allows you to build dashboards as well. If you’re just interested in only visualizing charts and trends, Plotly dashboards seem like the way to go. But to build reactivity into your dashboard (like subsetting your sample, changing date ranges, etc.), RShiny still seems like the more customizable solution.\n\n\nFinal thoughts\nOverall, it seems that ggplot2 is quicker to build and find what you want. With facet wrapping, the qplot() command, and ggsave(), you can whip something up fast. Plotly is better for dashboards, as you can interact with the plots. I feel like Plotly has a better syntax and documentation, and so it may be easier to get a basic plot to look how you want it to. But ggplot2 seems to have more advanced features, so if you want to get into refined customization, you may want to stick with ggplot2. They’re both great, and serve different purposes, but I’ll be using plotly for my RMarkdown and RShiny visualizations going forward.\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2017-10-19-used-cars-analysis/used-cars.html",
- "href": "posts/2017-10-19-used-cars-analysis/used-cars.html",
- "title": "Buying a used car the data science way: Part 2",
+ "objectID": "posts/2022-11-18-substance-over-style/index.html",
+ "href": "posts/2022-11-18-substance-over-style/index.html",
+ "title": "Resume Substance over Style",
"section": "",
- "text": "[Update 2021-11-16] This analysis was originally written on my old blog here. You can find the source code for it here.\nThis is part 2 out of a two-part series on scraping used car data. Check out part 1 to learn how to scrape the data.\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Earlier in my career I got feedback from a prospective recruiter telling me my resume was “visually imbalanced”. Of course, the recruiter wanted my business.\nBut I went on to get over 70 interviews with that same template. I got a job at FAANG and a top startup using that resume template. I haven’t changed it.\n\nTakeaway\nHow you write your resume and what’s on it will take you farther than how you design it.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2017-10-19-used-cars-analysis/used-cars.html#youre-in-the-market",
- "href": "posts/2017-10-19-used-cars-analysis/used-cars.html#youre-in-the-market",
- "title": "Buying a used car the data science way: Part 2",
- "section": "You’re in the market",
- "text": "You’re in the market\nSo you want to buy a car, but you don’t know anything about them? Welcome to my life.\nYou show up at the dealer and there’s a sticker on the window. You know the difference between make and model, but you soon learn what a trim is. Some versions come with leather. Some have a sun roof. Some have all wheel drive. Some have 20k in miles, and a similarly priced car in a higher trim is at 40k miles. How do you know what you’re really paying for, and what these features are worth? And how do you know what it’ll be worth when it’s four years older and has an additional 40k miles?\nPretty advanced questions for someone who just learned what a powertrain is.\nIn this analysis, I’ll dive into how you can use data to learn a lot about an industry in a short amount of time. I’ll first dive into knowing what you’re buying (now and later). Then, I’ll dive into how Truecar and others might be finding their ‘deals’."
+ "objectID": "posts/2022-11-18-writing-a-tech-resume/index.html",
+ "href": "posts/2022-11-18-writing-a-tech-resume/index.html",
+ "title": "Writing a Tech Resume",
+ "section": "",
+ "text": "TL;DR: Use the “Accomplished [X] as measured by [Y] by doing [Z]” format.\n\n\n\n\n\n\nWriting resume is painstaking, but let me help jumpstart you. My resume below has gotten me dozens of interviews with top tech companies (Apple, Google, Facebook, Amazon, Microsoft, Waymo, LinkedIn, Hopper, etc).\nIt got me in the door, but let me be clear I didn’t pass most of those interviews 😝. That’s the point of the resume: open the door.\nLet me help you get in the door.\n\nMy template\nHere’s the google doc link to my resume template, also posted below.\nhttps://docs.google.com/document/d/1n8W79UBiWYhMWHeF4sNDtwaWxySRwU3t9isodbUuycI/view\nYou can copy this template and put in your own content.\nI designed and wrote my resume the way I did by following the guidance of these resources: \n\nBlog: Google’s Accomplished X as measured by Y by doing Z formula\nYouTube Video: Google’s How to write a resume\nBlog: Cracking the Coding Interview: how to write a resume\n\n\n\nCommentary\n\nIt takes some practice, but the XYZ formula is very powerful.\nEmboldening words can be helpful to people who have 3 minutes to scan your resume.\nI’ve never gotten commentary on how long my resume is or crammed. I’m guessing people just glance for a few keywords and some decent content and assume the rest of the resume is good.\nI’ve had a lot of discussions in phone calls about my Personal Projects sections. Fellow techies all have their own personal projects and they want to know about mine. It sends a signal you’re intrinsically motivated.\nWhile my resume is long-ish, I try to think: what story does this bullet point tell? What story am I trying to convey?\n\n\n\nHow can I Help?\nI hope this helps! Send me a message at bryanwhitingcoaching@gmail.com if I can help review a resume or give more tips.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2017-10-19-used-cars-analysis/used-cars.html#what-am-i-paying-for",
- "href": "posts/2017-10-19-used-cars-analysis/used-cars.html#what-am-i-paying-for",
- "title": "Buying a used car the data science way: Part 2",
- "section": "What am I paying for?",
- "text": "What am I paying for?\nTo do any analysis, you need data. Please see my most recent post if you’re interested in learning how I got the data for this analysis. Suffice it to say, I webscraped it. The data is not my own, but belongs to Truecar.\nI love using linear regression to understand my data on a granular level. In my data set, I was able to extract the make, model, year, mileage, trim, and a flag for whether the car was all wheel drive (awd == 1) or not. Putting this data into a linear regression, we can see some interesting results. First let’s look at the data we have:\n\nlibrary(dplyr)\nload('../../data/used-car/tc-ford-edge.Rda')\n# Get the prices into the appropriate scale and remove some extraneous\ndf$price = df$price*1000\ndf %>% select(-stats, -url, -int, -vin, -v6, -trueprice, -resid) %>% head()\n\n## price year mileage trim awd location ext deal\n## 1 16000 2014 28.477 SE 1 Sterling, VA <NA> 6514\n## 2 13300 2013 64.315 SE 0 Chantilly, VA <NA> 1784\n## 3 13800 2013 72.111 SEL 0 Fairfax, VA Mineral Gray Metallic 3331\n## 4 16000 2014 27.490 SE 0 Manassas, VA Oxford White 5448\n## 5 12000 2013 66.825 SE 0 Clarksville, MD Mineral Gray Metallic NA\n## 6 14900 2014 65.157 SE 0 Manassas, VA Tuxedo Black Metallic 4436\n## mpyr\n## 1 7.12\n## 2 12.86\n## 3 14.42\n## 4 6.87\n## 5 13.37\n## 6 16.29\nNow let’s see the regression results.\n\nmod <- lm(price ~ as.factor(year) + mileage + trim + awd, data = df)\nsummary(mod)\n\n## \n## Call:\n## lm(formula = price ~ as.factor(year) + mileage + trim + awd, \n## data = df)\n## \n## Residuals:\n## Min 1Q Median 3Q Max \n## -4755.7 -854.6 -2.7 877.5 5024.2 \n## \n## Coefficients:\n## Estimate Std. Error t value Pr(>|t|) \n## (Intercept) 22216.400 788.885 28.162 < 2e-16 ***\n## as.factor(year)2013 1267.037 625.804 2.025 0.0442 * \n## as.factor(year)2014 3160.017 645.152 4.898 1.92e-06 ***\n## as.factor(year)2015 5664.470 710.322 7.975 9.56e-14 ***\n## as.factor(year)2016 6166.658 821.469 7.507 1.68e-12 ***\n## as.factor(year)2017 9082.971 878.956 10.334 < 2e-16 ***\n## mileage -91.951 5.625 -16.347 < 2e-16 ***\n## trimSE -4983.725 409.765 -12.162 < 2e-16 ***\n## trimSEL -1822.889 291.400 -6.256 2.17e-09 ***\n## trimSport 3419.270 374.977 9.119 < 2e-16 ***\n## trimTitanium 99.242 568.428 0.175 0.8616 \n## awd 1277.734 222.680 5.738 3.30e-08 ***\n## ---\n## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1\n## \n## Residual standard error: 1450 on 211 degrees of freedom\n## Multiple R-squared: 0.9231, Adjusted R-squared: 0.9191 \n## F-statistic: 230.2 on 11 and 211 DF, p-value: < 2.2e-16"
+ "objectID": "posts/2022-06-23-new-ideas-are-distractingly-exciting/index.html",
+ "href": "posts/2022-06-23-new-ideas-are-distractingly-exciting/index.html",
+ "title": "New ideas are distractingly exciting",
+ "section": "",
+ "text": "When a new idea comes to my mind, I want to act on it right away. Perhaps it’s an exciting analysis idea, a new business strategy I want to scope out, or a question to ask a team member.\nBut as I look back at my backlog of things to work on, all of these ideas were also equally exciting.\nWhy is it that new ideas have more weight than old ideas?\nThey say that effectiveness is working quickly on tasks. Efficiency is knowing what to work on. Discipline is the ability to overcome the excitement of a new idea and place it in proper context of all ideas. Sanity is actively saying no to many ideas so you don’t get inundated.\nPerhaps a good strategy is to have a “gut check” for ideas. If they don’t pass the highest threshold then let them go.\nProcess:\n\nWhat’s the new idea?\nWhat’s the potential impact? What would I do with the output of this idea?\nHow feasible is it?\n\nStep 3 is tricky because I don’t want a good idea to go to waste just because it’s hard. Judgment is weighing the trade off between impact and feasibility.\nI like Google Forms for logging ideas. It’s easy to save a Google Form URL to the home screen of my iPhone to make logging new ideas easy. Grooming those ideas is important. Letting go of old ideas is essential. If only I could do the essential…\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2017-10-19-used-cars-analysis/used-cars.html#interpretation-matters",
- "href": "posts/2017-10-19-used-cars-analysis/used-cars.html#interpretation-matters",
- "title": "Buying a used car the data science way: Part 2",
- "section": "Interpretation matters",
- "text": "Interpretation matters\nFirst, we see the (Intercept). This is the average value of a car for every level of every feature not show. So it represents the average price of a year 2000 with 0 miles, Limited trim, without all wheel drive. This allows us to now see what the average effect of, say, year is, holding all else constant. Every unit is in terms of dollars, so we can see that for as.factor(year)2013 that 2013 cars are worth $1,267 more than 2012 cars. This makes sense.\nI scaled mileage to be in the thousands so that it’s a little easier to interpret. So we see that for every 1,000 miles you drive a Ford Edge, it decreases the value of the car by $91 dollars. This tells you a few things. One, if I’m being offered two cars that are 10K miles different, there should be about a ~$900 difference in the car.\nSecond, if I’m planning on driving my car 30k miles over the next two years, how much should I expect the price to drop just due to mileage? Well, about -$91 * 30 = -$2,730. And maybe if you bought a 2014 model, two years later it could perhaps be like owning a 2012 model today. What’s the two-year difference between a 2014 and a 2012 model? About $3,160 dollars."
+ "objectID": "posts/2020-12-15-calendaring-todos/index.html",
+ "href": "posts/2020-12-15-calendaring-todos/index.html",
+ "title": "Get rid of the todo list. Calendar everything.",
+ "section": "",
+ "text": "I’ve had less meetings and more unstructured time during WFH. It can be isolating, which can slow down productivity.\nHere’s a tip I read today on how to be productive with unstructured time: put your to-do list directly into a calendar.\nThat way, you’ll always have time for the important things. Remove the things that take up your time and aren’t as important. I like it.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2017-10-19-used-cars-analysis/used-cars.html#inference",
- "href": "posts/2017-10-19-used-cars-analysis/used-cars.html#inference",
- "title": "Buying a used car the data science way: Part 2",
- "section": "Inference",
- "text": "Inference\nNow it’s getting interesting! You’re now starting to remember how your mom told you to never buy a car new, because once you drive it off the lot, it’s worth significantly less. Well, we can see that if you own a brand-new 2017 model and a 2016 model, they’re about $9082 - $6166 = $2916 different. So yea, just by getting the 2017 model you’ll lose $3,000 in the first year. See how 2016 is only $502 dollars more than the 2015? Having the 2015 model is practically the same price as a 2016. So you might as well buy the 2016 model and save a year of wear.\nWell, my theory kind of starts breaking down because 2015 cars and 2014 cars are similarly spaced as 2014 and 2013 ones. So maybe I shouldn’t read too much into this. But the ability to see these trends shouldn’t stop you from asking important questions, like, why is there a difference between years?\nWhen investigating the Nissan Murano, I observed a $5430 difference between 2015 and 2014. That’s non-trivial considering the other years were equally spaced. Doing a little research I saw that Nissan changed the 2015 Murano style, and the new style was apparently worth a lot more. Turns out the same thing happened here for the Ford Edge - the 2015 model is slightly longer with redesigned interior.\nYou can also use this framework to get a decent sense for what features are important to you. Like, is having all wheel drive really worth $1,277? If you live in an area where it snows, it probably is. But if you live in Florida, you may not need it.\nNot only is this framework helpful for comparing within-make-model differences, it helps you get a sense of between-make-model differences. Say, comparing the Ford Edge with the Nissan Murano and the Toyota RAV4. The coefficient for milage on the Edge is -$91. On the Murano it’s -$83, and on the RAV4 -$61. Remember this is the cost per 1,000 miles. That tells me that if I ‘spend’ 30k miles on a car over two years, I’ll lose -$2,730 on the Edge, but only -$1,830 on the RAV4, saving me $900. This also validates that Toyotas hold their value more."
+ "objectID": "posts/2016-11-07-starting-with-jekyll/starting-with-jekyll.html",
+ "href": "posts/2016-11-07-starting-with-jekyll/starting-with-jekyll.html",
+ "title": "Starting with Jekyll",
+ "section": "",
+ "text": "Hello world.\nI wanted to start a blog. I wanted to set it up for free, use a custom URL (and not something.bloggingplatform.com), and be able to both blog and create tutorials. I didn’t mind it possibly being technical1. Enter Jekyll.\n1 Specifically, I was interested in being able to use the command line for my posts. Additionally, I didn’t mind seeing or using a little bit of code.If you want to get your blog in 10 minutes, skip to below.\n\nWhy Blog with Jekyll?\nHere’s why you can and should blog with Jekyll (if you’re a data scientist):\n\nJekyll has beautiful, free, open-source templates.\nGitHub will host your website for free2.\nJekyll is static.\nJekyll can be simple to set up3.\nI tried Wordpress and blogger.com and got frustrated.\nThe blogging content you create is very portable (easy to take your posts from one blog site to another).\nCreating blogs in Markdown syntax is a dream (even for non-technical folks).\nYou can manage your blog from your terminal\nYou can manage your blog from your phone (if you have a GitHub-editing app)\n\n2 Someone has to host it (i.e., store and display all the files). I started off trying to set up a WordPress blog, but you have to pay another company to host. I didn’t want to bother.3 If you try to set it up by following the instructions on jekyllrb.com, that’s the complicated way. That’s how I originally started, got lost for five hours, and stopped. Six months later, I found a better way.But if you don’t like debugging occasional errors, you probably shouldn’t blog with Jekyll.\nWhat is Jekyll, you ask? Jekyll is an open-source blogging platform. Anyone can write a template and post it, and anyone else can use that template. Perusing Jekyllthemes.org, you can pick out any template you’d like.\n\n\nHow did I get started with Jekyll?\nAs a blogging noob, I’ve discovered there are hundreds of ways to generate a blog (e.g., WordPress, Squarespace, blogger.com, Weebly, etc.). But when my colleague Arthur Lui showed me his blog, I wanted something similar. I first tried Wordpress, then tried blogspot, then gave up and followed Arthur’s example and landed with Jekyll.\n\n\nSet up a blog in 10 minutes!\nI followed this Jekyll tutorial to set up my first blog. I’ll simplify those steps here to get you up and running quickly:\n\nGet a GitHub account. (2 min)\nGo to Jekyllthemes.org. All of these blog themes are free. Find one you like, and Fork the repo (i.e., “repository”). (2-60 min, depending on how long you peruse)\nIn GitHub, rename the newly-forked repo from “REPO name” to “yourgithubusername.github.io”. (1 min)\nNow go to “githubusername.github.io”. Voila, you have a blog.\n\nAnd if you want to start writing posts, go to your “_posts” folder. Then create new files and follow the default templates that the repository provides. (The blogs will need to be written in Markdown syntax.)\nThat was my 10 minute promise. If you feel a bit cheated or lost, go to the well-done tutorial and do all the steps. But the steps I listed are the essential steps to setting up a blog from scratch. So if you want to make a few blogs really quickly, after you figure out the first one the others follow nicely. For example, once I finished the tutorial, I followed those steps above to try two other themes before deciding on the Centrarium theme.\nTo get Centrarium to work, I did have to change some configurations in the “_config.yml”, so there can be some minor debugging depending on the theme you choose. But with a little patience and some Googling, you’ll get it.\n\n\nAdd to Jekyll a custom URL from GoDaddy\nYou’ll notice I have a custom URL and not “bryanwhiting.github.io”. I used GoDaddy to get a domain. (I have no affiliation with GoDaddy.) Here’s how I linked GoDaddy with my GitHub Jekyll blog:\n\nGet a domain.\nIn your Github repo, change (or create) a CNAME file in the main directory. (See my CNAME file for an example). And for the CNAME file contents, just type in the new “www” that you registered through GoDaddy.\nThen, on GoDaddy, go to the section “DNS Management”. Here’s where you tell GoDaddy how to connect your newly purchased “www” with GitHub. Create the following connections:\n[Type, Name, Value, TTL] = [A, @, 192.30.252.153, 600 seconds]\n[Type, Name, Value, TTL] = [A, @, 192.30.252.154, 600 seconds]\n[Type, Name, Value, TTL] = [CNAME, ftp, yourgithubusername.github.io, 600 seconds]\n[Type, Name, Value, TTL] = [CNAME, www, yourgithubusername.github.io, 600 seconds]\n\nIf you have the CNAME file in your GitHub folder, GitHub and GoDaddy will talk and after a few minutes your custom “www” will work. You don’t technically need the two [A] steps, but those connections allow users to type “bryanwhiting.com” instead of “www.bryanwhiting.com”. It can take up to 24 hours before these [A] connections enable, so don’t be worried if the short URL doesn’t work first try. But the full URL should work soon enough.\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2017-10-19-used-cars-analysis/used-cars.html#identify-a-good-market-deal-and-how-truecar-might-do-it",
- "href": "posts/2017-10-19-used-cars-analysis/used-cars.html#identify-a-good-market-deal-and-how-truecar-might-do-it",
- "title": "Buying a used car the data science way: Part 2",
- "section": "Identify a good market deal (and how TrueCar might do it)",
- "text": "Identify a good market deal (and how TrueCar might do it)\nBeyond the research above, you can use such a model to identify deals in the market and guide your buying decision. Here’s how:\n\nMake a PriceFinder: Get the residuals (actual minus predicted value) for each car. The more negative the residual, the more this car is a good price! This may be how TrueCar and CarGurus get their ‘deals’. They fit a model (probably one better than what I’ve created here) and use its residuals to value each car.\nIdentify ‘overpriced’ cars and try to possibly haggle those dealers down. Again, use the residuals here.\nUse this model to predict ‘out-of-sample’ cars. You can use this model built on TrueCar data and find a car on the classifieds or another site and evaluate whether it’s a good deal or not.\n\nHere’s an example of how we do that. First, let’s predict a hypothetical 2015 SEL Ford Edge FWD with 27.6k miles:\n\nthat_car <- as.data.frame(list(\n year = 2015,\n mileage = 27.6,\n trim = 'SEL',\n awd = 0\n))\n\npredict(mod, newdata = that_car)\n\n## 1 \n## 23520.13\nWe’d expect, on average, such a car to cost $23,520.13. Is what the dealer’s offering you above or below that? If they’re offering $25,000 then you know the car is $1,479.87 higher than what you might expect."
+ "objectID": "posts/2021-07-17-authors-angle-matters/index.html",
+ "href": "posts/2021-07-17-authors-angle-matters/index.html",
+ "title": "The Author’s Angle Matters",
+ "section": "",
+ "text": "Imagine the following story:\n\n“In recent events, a ferocious fox savagely attacked a Turkey, who fortunately was able to escape to safety by pecking out the fox’s eyes.”\n\nSimple story. A bit gruesome. But, what in today’s news cycle is positive? How can we be better readers of the news when it’s targeted at making us angry?\nFoxes and turkeys are highlighted in the news all the time and it’s often the worst of humanity. The goal of many authors is to evict emotion from you so you’ll keep reading. Anger. Fear. Worry. Those keep you hooked.\nWhat’s the author’s motive? Often we have to guess it. Wouldn’t it be nice instead if authors were transparent about why they were writing a story:\n\nAuthor 1: I’m a Turkey conservationist with a passion for protecting turkeys from the rampant fox problem we have. Growing up, a Turkey saved my life when I was drowning in a lake.\nAuthor 2: I’m a narcissist who hates foxes because one time they ate my cat. I really just need therapy, but I can’t fathom talking to anyone about my problems. I want you to be as angry as I am.\nAuthor 3: I’m here to get promoted and if my boss sees that I have a 10% increase in viewership then our company’s bottom line increases. Im an opportunist who will say anything and take any side that will maximize company profits. How do these three perspectives change the storyline?\n\nWouldn’t it be great if we could know the author in addition to what the author is writing about? Every article should have not just “about the author” but, “three people’s perspectives about the author’s intent in writing this that know the author well.” Of course, you’d need to verify the three verifiers, but it’d be a start\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2017-10-19-used-cars-analysis/used-cars.html#how-to-improve-our-model",
- "href": "posts/2017-10-19-used-cars-analysis/used-cars.html#how-to-improve-our-model",
- "title": "Buying a used car the data science way: Part 2",
- "section": "How to improve our model",
- "text": "How to improve our model\nNow, this is a pretty naive model, but it’s a model. As George Box said, “All models are wrong, but some are useful.” But here are some quick ways to improve this model:\n\nCheck for interactions. An interaction between miles and year would tell you whether the effect of miles would change across different years. For example, I think adding 10k miles to a 2017 would devalue the car more steeply than adding 10k miles to a 2012 model. During my quick research I checked for interactions, but I didn’t find many to be significant and I chose to ignore them for simplicity.\nUse a linear trend for year rather than a ‘factor’ trend. I’ve chosen here to treat year as a factor, which implies that I think that each year is completely different. Had I used year as a numeric feature, I could have possibly interpreted year as, “Each additional year increases/decreases the price by X on average, holding all else constant.” I liked the simplicity of thinking that each year was different, and since cars usually update models every four years, I wanted to be able to see where this happened. You couldn’t easily see this if year were a linear trend. Objectively, what would be a good way to choose? Fit both models and see which had better R-squared or AIC.\nDo model diagnostics. Check the residual plot to make sure the errors are normally distributed. Check for outliers. Each of these affects the accuracy of your residuals. I did some brief checks and was genearlly satisfied."
+ "objectID": "gratitude/2023-12-26-i'm-grateful-for-blankets/index.html",
+ "href": "gratitude/2023-12-26-i'm-grateful-for-blankets/index.html",
+ "title": "I’m grateful for blankets",
+ "section": "",
+ "text": "A blanket\n\n\nI’m outside right now enjoying the warmth of a blanket. It’s a special thing. It’s warm. It’s fuzzy. And the only question I have is: is it warm because it’s fuzzy? Or is it fuzzy because it’s warm?\nIn the Hitchhiker’s Guide to the Galaxy the most important item any many could carry was a towel. It’s a multifaceted tool. Perhaps a blanket is the towel’s cousin.\nMy grandma loved to quilt, so me and all my family have hand crafted blankets. Even great grandchildren have blankets long after my grandma deceased because she had a surplus in her basement.\nThere’s something special about being wrapped in something warm, especially if it’s made by someone who loves you. Kinda sends a special hug in a way that a towel just can’t.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2017-10-19-used-cars-analysis/used-cars.html#takeaways",
- "href": "posts/2017-10-19-used-cars-analysis/used-cars.html#takeaways",
- "title": "Buying a used car the data science way: Part 2",
- "section": "Takeaways",
- "text": "Takeaways\nSo, a data scientist walks into a dealership with a computer with a registry of thousands of used cars on it. He negotiates with the front-office salesman. And the salesman says…\n\nNon-statistical considerations\nWhen you’re buying a car from a dealership, remember that it’s their job to sell you that car. Here are other random things I learned to think about during the car-buying process.\n\nHow many miles has it been driven, and how old is it? (We found a car with 45,000 miles in one year. That’s almost two trips around the entire world in a year! This is also 3750 miles per month (a cross-country US trip), or about 125 miles per day. What kind of driving do you have to do in a year to get that?\nHow many prior owners were there?\nWas the car previously a rental? A lease? A fleet?\nRental: Abused by X number of people? But also perhaps better maintained by rental companies?\nLease: What kind of person would lease?\nFleet: Probably a corporate car. Like a rental, but different kind of drivers.\nWhere did the car come from?\nIs there snow where they live? And do they use salt on the roads there? (Salt = rust underneath)\nWhat else does the geography tell you about the possible wear on the car (causing long-term problems)\nWhat does the CarFax or Autocheck say?\nWas the car sold at auction? When? What could the time between sales tell you about the car?\nIf someone gets into an accident and doesn’t report it, the CarFax won’t tell you. I drove 45 minutes to look at a ‘great’ deal only to learn the door was nearly falling off.\nGood luck haggling the dealer down. Car dealerships don’t usually make that much on the car, but make their money on warranties and fees. Your best chance is to haggle on the deals being offered (terms and agreements).\nDon’t get the powertrain warranty. This is like the “french fries” of the auto industry. Pure profit.\nEvery dealership has quota. Going in at the end of the month might help you, but the dealership will be packed.\nUsed cars are bought at auction. Carmax buys all their cars at auction and only keeps 1/3 of them. Local dealers also get their cars this way, and they don’t always know what the car looks like till it gets to their lot. Check the Carfax to see if any prior dealers ‘flipped’ the car quickly.\n\nUsed cars have a shelf life. If a car isn’t sold in 45-60 days, some dealerships might have to auction off the car. That means you’ll see a tiering in the price. Less-favorable cars will see the price drop several times in its life cycle. Check CarGurus to see how many times the price has dropped, and try to guess for yourself if it might drop again (or just get sold). They know their prices. They’ve got their back-office guys looking at every price of every same car within 100 mile radius. While I’m also able to get as many of those cars as available on TrueCar, I’m not able to segment by all the various features. It’s easy to webscrape for Make, Model, Year, AWD/FWD, and maybe V6 engine, but it’s very hard to get the details like leather, sunroof, twin-turbo, etc. That’s where the pricing gets really interesting.\n\n\nFinal thoughts\nUltimately, a car is bought by irrational humans, and sold by more experienced ones too. So all of this data will probably go out the door when you try to make a deal.\nSo a data scientist walked into several dealerships, and left frustrated because nobody budged on their prices. Doesn’t matter how much data I have, or what I think I know, everyone’s got their price."
+ "objectID": "gratitude/2023-12-24-light/index.html",
+ "href": "gratitude/2023-12-24-light/index.html",
+ "title": "Light",
+ "section": "",
+ "text": "I installed 72 lights today. This took drilling a hole into my fence, then making the cord taught, then screwing the screw into the cord’s loopy thingy. Then I move on to the next one.\nThe whole process took three hours. But I’m glad it took three hours. I appreciate it so much more because I did it myself. I thought carefully about each hole I drilled.\nAnd it felt good to sit out there and listen to an audio book (lots of Dune today, some Killers of the Flower Moon).\nI’m grateful for Thomas Edison. For the inspiration he received, for the hard work he did to give the gift of light to the world. And if he hadn’t done it someone else would have, which is also something to be grateful for. There are a lot of smart, good people on this earth.\n\n\n\nNot what it looks like, but how it feels\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2020-12-15-data-beats-opinions/index.html",
- "href": "posts/2020-12-15-data-beats-opinions/index.html",
- "title": "Data Beats Opinions",
+ "objectID": "lists.html",
+ "href": "lists.html",
+ "title": "Lists of Things",
"section": "",
- "text": "When I co-started a company several years ago, my founders and I would canvass neighborhoods knocking doors. We’d discuss our ideas for a new product one door at a time. After each conversation, the product took a new form. Eventually, the product took shape and we had a successful kickstarter launch.\nSince that time, I’ve thought about how valuable it was to get information straight from the source.\nAlberto Savoia’s new book, “The Right It”, is re-teaching me this principle. Data beats opinions, he says. In another book I’m reading, “Trustworthy Online Controlled Experiments”, the former CEO of Netscape is quoted saying,\n\nIf we have data, let’s go with data. If all we have are opinions, let’s go with mine. - Jim Barksdale\n\nData sourced from your manager, your stakeholder, or your customer are invaluable to help you know if you’re solving their problem. Too often it’s easy to get stuck in the original solution that comes to our mind.\nData from customers directly shape the product. Data sourced at scale from online experiments refine it.\nOpinions from the customer shape the product. Their behavior (captured as data) speaks louder than their words. Frequently, it’s challenging to correlate the two. We’re left to guess why someone actually clicked, bought, churned. It’s a beautiful cycle.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Ongoing bookmarks and lists of things I want to remember.\n\n\n\n\n\n\n\n\n\n\n\n\nCool AI Tweets\n\n\nRunning list of cool things I’ve seen.\n\n\n\n\n\n\nDec 14, 2023\n\n\n\n\n\n\n\n\nNo matching items\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only.",
+ "crumbs": [
+ "Lists of Things"
+ ]
},
{
- "objectID": "posts/2022-12-06-more-new-businesses-than-ever/index.html",
- "href": "posts/2022-12-06-more-new-businesses-than-ever/index.html",
- "title": "Models as a Service and the Future of Data Science",
+ "objectID": "christianity/christmas-empty-tomb/index.html",
+ "href": "christianity/christmas-empty-tomb/index.html",
+ "title": "Post Christmas blues? Remember the empty tombs",
"section": "",
- "text": "Businesses are Launching Like Crazy\nThere’s never been a better time to start a business. The Census shows there are over 200,000 new business applications per month more than pre pandemic.\n\nWhat’s driving this?\nListening to The Unintended Consequences of Working from Home it dawned on me that there’s a hyper acceleration of innovation happening.\nConsider the rise of AI tools like ChatGPT and StableDiffusion. Chamath Palihapitiya discusses the expansive applications for these services. What was once SaaS will become MaaS (models as a service). There will be models for everything: copy writing, code debugging, converting wireframes into working apps, ad generation, book writing…\nWhile tools like ChatGPT are just in demo mode, it’s wildly apparent they’ll be an impetus for new ideas and reducing the friction of getting to the answer. For example, as a data scientist if I need boilerplate code to clean some data, I can ask ChatGPT for the 80% solution. I’m already spending 30% of my day copying code snippets from StackOverflow and making it work for my problem. ChatGPT will just accelerate that process. 10 years from now if Neuralink can read my thoughts I don’t even need a keyboard to do this…we’ll build things at the speed of thought.\nConsider this for writers and marketers. They’ll use ChatGPT to unblock them, generate ideas, and get started. Innovation across the board will accelerate because the kinetic friction to getting started will reduce to zero. You can then spend your time on the editing and refining.\n\n\nThe Best Time to Start a Business\nI digress. Why is now the best time to build a software business?\n\nCompute is cheap. Gone are the days of having to build a server to prototype and idea.\nBarriers to prototyping are nearly free and scaling with the cloud is nearly infinite. Consider how the Gas App can launch to 100,000 new users per day with just a few developers.\nModeling is now easy. All cloud services have their own AutoML tools to build machine learning models for you.\n\nThe center of all of this is scalable data infrastructure and AI.\n\n\nFuture of Data Science\nThe data scientist of today may still be using sklearn to manually do hyperparameter tuning, but the data scientist of tomorrow will either be exclusively building MaaS B2B products or be the one person in their organization who will leverage all the MaaS tools.\nIf we can ask ChatGPT for complex answers it learned from the internet, why couldn’t something similar be trained on your company’s internal data? If that’s possible, then why couldn’t you just have a Q&A session with your internal ChatGPT? Consider the following example:\n\n“Hey KITT, what are the user segments generating the most revenue on Tuesdays at 5pm?” (Analytics)\n“How many sales do we anticipate from these users next Tuesday?” (Machine Learning)\n“What ads have been most effective for us in driving new sales in the past 7 days?” (Statistics)\n“Okay, KITT, generate 3 variants of that ad for next Tuesday” (StableDiffusion generating a new Ad)\n“I like ads A, B and C. launch an experiment on Facebook/TikTok, run it for three days and dynamically allocate ad spend to the best ad.” (Business integrations, statistics, ML models)\n\nThe data scientist of the future will be the one with deep understanding of these tools and an awareness of their biases. It will be very difficult to know when they’re lying to you. The in-house data scientist will be keen on knowing the biases in the data fed into these algorithms and how to make judgments on where to go next.\nEarly stage companies won’t need data scientists because most of this stuff will be off the shelf and easy for an engineer to integrate. That means that the data scientists will join these MaaS companies to build out optimized products that solve these niche problems.\n\n\nTakeaways\nData Scientists shouldn’t be scared their jobs will disappear. They should be excited that they’ll be joining companies that automate what they’re currently doing and offer it as a service.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "An empty spot in the room\nJust cleaned up the Christmas tree and all its ornaments and needles. The glittering, shimmering, and sparkling lights that have warmed us for the last month are officially gone. Packed up. Shipped out. Sent home.\nMy child remarked, “it feels so empty without the tree!” I couldn’t help but agree.\nThe thought dawned on me, perhaps this is what the disciples felt once His light left the earth. Some accounts declare there was thundering and lightning and three days of darkness that covered the earth after He gave His life.\n\n\n\nA dead Christmas tree reminds us of a Christ who died for us on a tree, the cross.\n\n\n\n\nAn empty room, an empty tomb\nBut just as His easily preventable departure left friends and parents weeping (because they didn’t fully understand what He was dying for, (Matthew 16:21-23), perhaps we can remember a little of that sadness that they felt as we clean up the tree. The child was born. He was born to break the bonds of death. He died, some describing He died on a tree. And now, my pine tree representing Christ(mas) is also dead.\n\nWho his own self bare our sins in his own body on the tree, that we, being dead to sins, should live unto righteousness: by whose stripes ye were healed. (1 Peter 2:24)\n\nPerhaps it’s also symbolic we cut down the tree to put it in our homes. It has an expiration date, just like Christ’s life. He knew it was time (John 13:26-27), even if nobody else could comprehend.\nBut remember what happened after? He resurrected. He brought His light back to His friends.\nPerhaps New Year’s lights can remind us of His return. And perhaps a New Year’s resolution can remind us that He invited His disciples to change their ways and go and preach to the gentiles (people previously not allowed to have His gospel and His gift of baptism and the Holy Ghost), to share the gift of Christ(mas) with everyone.\nThe death of Christ marked the fulfillment of prophecy and the fulfillment of the Law of Moses, which meant that the Jews and the house of Israel were no longer the only ones chosen by God but now everybody could receive the gospel (Galatians 3:29).\nLet’s take this new year to remember all are loved. All are remembered. All are invited to happiness and peace through Christ.\n\n\n\nDead tree? Remember the living Christ!\n\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2022-06-20-blogging-with-quarto/index.html",
- "href": "posts/2022-06-20-blogging-with-quarto/index.html",
- "title": "Blogging with Quarto, and why I don’t use Medium anymore",
+ "objectID": "christianity/unknown-god/index.html",
+ "href": "christianity/unknown-god/index.html",
+ "title": "An Unknown God",
"section": "",
- "text": "Been here before, eh?\nI probably shouldn’t write this post considering I’ve also written posts “blogging with Jekyll”, “Blogdown”, and “distill” before. I’m flaky, what can I say. I’ve changed my blogging framework 7 times in 5 years (I started with blogger.com and Wordpress before Jekyll, and even used Medium for 25 posts.)\nBut it’s not my fault. There are too many static site generators and it seems that Rstudio is as flaky as me. Meaning, they built rmarkdown, created hugodown, blowdown, bookdown, and distill. Hugodown is a lightweight version of blogdown, and I never figured out why it exists. Then there’s workflowr and so many other great, random products out there.\nWhy quarto for me? It seems to have all the stuff that distill has, and then some. And given its design and collaboration with Wes McKinney (pandas guy) in building quarto, and given that Wes wrote his latest book with it, I think this will be here to stay for at least 5 years.\n\n\nWhy do I avoid Medium\nWhy don’t I blog with Medium? I wrote 25 posts there over 9 months and loved it. It makes writing and publishing incredibly easy, which is why I did it. With distill/blogdown the limiting factor was I’d have to be at my computer to get a post in. I love writing on my phone - I have over 2000 journal entries with Day One over seven years, several of which are “draft” posts for a blog that won’t see the light of day since I don’t care about them after I’ve written it. I want to be able to blog frequently, a short little blurb to keep up my writing, and medium enabled this.\nBut 1) I didn’t get many views (not that I care) and 2) my writing is stuck on their platform (confirmed to me this week as I’ve tried several old, broken tools that attempt to export posts out of medium).\nI got 50x the views on my blogdown site just from organic google searching. From reading other people’s posts about leaving Medium, it’s seems that most traffic to Medium posts comes from outside of Medium.\nSo while I’m certainly not into blogging for the money, I wouldn’t mind getting the content to people who it can help. And if medium didn’t enable this, what is its purpose?\nSo I’m back and finally decided that I’m going to own my own platform. Seth Godin wrote a blog about this topic once. Why be on Medium, substack, Facebook, etc or any other up an coming platform? We’ll they’re all skewed toward the 1% of people who make money on the platform. Everyone else is wasting their time. So might as well own your content. And better to focus on writing than waste time catching up on every platform.\nAlso, it’s fun! There’s a fuzzy feeling to blogging.\n\n\nFuture of quarto?\nI’m speculating here, but Quarto can be used withR Studio, Jupiter or VS Code. The last editor seems excitingly suspicious.\nPerhaps R Studio sees the writing on the wall that VS Code is taking over the IDE world and perhaps taking over the long term future of R Studio IDE. I personally don’t see why anyone would pay for R Studio when VS Code is free and has a better interface. And while R Studio has a few better integrations for R at the moment, it won’t once enough people see the light for VS Code and turn their contributions there. I’ve been using R on VSCode for two years now and it rivals R Studio in 95% of the feature set, but 100% has better window management.\nSo Quarto seems like a great opportunity for R Studio to begin monetizing on a publishing platform like R Studio connect. Platforms like Notion and Confluence are great for everyone except data scientists trying to share reproducible research. But a quarto server seems like a great way to finally bring in Python data scientists into the markdown-flavored documentation world. (Man, I personally can’t use Jupyter/Google colab notebooks when I can use R Markdown!) I’d heartedly welcome this prospect, though I have no idea what their plans for Quarto really are.\nTill next blog post about what blogging tool I’m going to use…\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "This post speaks to how I manage spiritual doubts, such as whether God exists. It got me thinking after reading a few religious posts by Rob Hybdman, someone whose professional (data science) work I greatly admire.\nHere’s a summary of some points I try to make here:\n\nHe doesn’t try to convince us to believe Him without a doubt, He asks us to follow him\nPeople who saw Christ didn’t even recognize Him as the Savior\nChristianity is hard to grasp and won’t make total sense when you think deeply about it.\nOur job is to keep learning, and ask Him to teach us about His nature\n\n\n\n\nI think too many people grow up thinking of God in black/white terms and when one, linking piece of the puzzle doesn’t fit or is missing, some throw Him out all together. Here’s my take on why people misrepresent God, and how this can lead to personal doubts of faith in Him and His Son.\n\n\n\nChrist asking his disciples, “Whom say he that I am?”\n\n\n\nAn unknown God\nA certain philosopher named Korihor once reasoned that people should stop paying tithing or following restrictive religious commandments because they’re wasting their time. Here’s his thinking:\n\nYea, they durst not make use of that which is their own lest they should offend their priests, who do yoke them according to their desires, and have brought them to believe, ==by their traditions and their dreams and their whims and their visions== and their ==pretended mysteries==, that they should, if they did not do according to their words, ==offend some unknown being, who they say is God==—a being who never has been seen or known, who never was nor ever will be. - Alma 30:28\n\nI love this because the argument of whether God exists or if Christ was the Savior is nothing new. (And many people I highly regard and esteem are tackling this “new” issue head on, as many in the world are questioning their faith.) But a lot of the same arguments that have existed continue to exist: why believe?\n…It’s not like anyone’s ever seen God anyway.\nLet’s assume for a minute that they have seen God. They saw His Son, His mortal example, walk the earth (John 5:19).\nWhile with His disciples, Christ himself took a moment to reflect on who people thought he was, and whether people actually recognized Him for who He is:\n\nWhen Jesus came into the coasts of Cæsarea Philippi, he asked his disciples, saying, ==Whom do men say that I the Son of man am?== And they said, Some say that thou art John the Baptist: some, Elias; and others, Jeremias, or one of the prophets. Matthews 16:13-14\n\nPeople had no idea who Christ was even though I’d bet either He or His disciples them His name many times while He was healing them.\nPeople are funny like that.\nConsider your own life. Have you ever misinterpreted a situation, even as it’s being explained to you?\nLet’s consider a magician who does a magic trick for you. She doesn’t show you how it’s done, but she does explain it to you with only words. Since you don’t specifically see how the magic trick is done, perhaps your mind wanders and you doubt what really happened.\nDo you trust the Magician? When He tells you who He is, after healing you, do you believe Him? When an unexplainable coincidence happens in your life, what do you chalk it up to?\n\n“Nah, that’s so highly improbable it must just be luck”… (whatever luck is…)\n\nI can’t imagine how it must have felt for people around Christ to be healed by rubbing mud on their eyes or washing in a river a few times or, or, … I can only honestly believe these people felt absolute wonder to the point of mental disassociation. How else do you mentally cope with being a paralytic your whole life and then you’re able to walk?\nIt’s entirely reasonable, then, that they just figured He must be like the other miracle workers they knew - Elias, John the Baptist, etc. They try to frame Him in their believable mental reference, i.e., “luck”.\n\n\nA strange God\nPerhaps though we can relate to when Paul, a lifelong Non-Christian and Christian persecuter, preached to his fellow Gentiles:\n\n16 Now while Paul waited for them at Athens, his spirit was stirred in him, when he saw the city wholly given to idolatry. 17 Therefore ==disputed he in the synagogue with the Jews==, and with the devout persons, and in the market daily with them that met with him. 18 ==Then certain philosophers== of the Epicureans, and of the Stoicks, encountered him. And some said, ==What will this babbler say?== other some, He seemeth to be a setter forth of strange gods: because he preached unto them Jesus, and the resurrection. Acts 17:16-18\n\nPaul got them. He was once like them: a gentile. An unbeliever. A disbeliever. But he eventually was converted to knowing Christ, and, as such, appeared to seem like a babbler when trying to talk to people who once were like him. They considered him and his teachings strange:\n\n19 And they took him, and brought him unto Areopagus, saying, ==May we know what this new doctrine==, whereof thou speakest, is? 20 For ==thou bringest certain strange things to our ears==: we would know therefore what these things mean. Acts 16:19-20\n\nMeeting Christ can be strange.\nUnderstanding him, so too.\n\n\nPutting a box on God\nConsider the question, “why does God let good people suffer?”\n(Other important questions like, “why did God encourage Abraham/Isaac/Israel to be polygamists but Adam not?”, or “why did God restrict the entire earth from receiving his Gospel until after Christ died, commanding His people not to marry or mingle with them but then change his tone ~33 AD?”, are worth similar consideration.)\nIt’s a false dichotomy: if God exists and loves us, He surely wouldn’t let us suffer. If suffering exists, especially of good1 people, God must not then exist.\n1 What is Good without God?2 The world is filled with unjust, seemingly unnecessary suffering. Consider innocent children starving across the world, being bombed in Gaza/raped in Israel, bombed in Ukraine, Yemen, gassed in Germany in the Holocaust, starved in Stalin’s camps, nuked by the United States in Japan, etc, etc, etc. (I sincerely don’t write this easily, nor to make light of the suffering.) all of the suffering is horrible. But it doesn’t for a moment mean God doesn’t exist. It actually affirms His existence, because if evil exists the opposite of evil must too.3 I don’t mean that in the Marxist political sense, because who’s to declare who’s suffering at any given moment? Is an affluent child born with cancer not a sufferer despite them not being poor?Consider all the suffering of innocent children2. The reality of human nature is sufferers vs non sufferers3. Any suffering is painful to bear, and nobody better understands this than Christ himself (D&C 19:16-19).\nThe point is, we can falsely equivocate that because God loves us and He had the power to stop suffering, He should stop it.\nDo you do this with any other relationship? Consider the hypothetical:\n\nFather Joe loves his children. But Joe let his son Bob get hit with a toy by little sister Susie. Joe let Susie suffer. Therefore, Joe acted in an evil and therefore does not exist.\n\nThis line of thinking is absurd to anyone with a brain, let alone to anyone who is a parent. They understand that sufferig is part of life, can be good for teaching, and that it doesn’t mean the parent doesn’t exist. Yet sometimes we wonder if God exists because we’re suffering.\n\n\nConsider alternative viewpoints\nTo be comfortable with unanswered questions from a yet unknown God, we should consider how the question is already answered.\nFor example, did Hod stop Christ from suffering? No. Did Christ stop himself from suffering on the cross when He had the power to turn water to wine for the thirsty, heal the suffering swordsman’s ear, or even raise Himself from the dead?\nI’m trying to rally around the point: we set up for ourselves false philosophical arguments about who or what God is and should be like when there’s a lot of nuance to be considered.\nIn my view, He’s more of a Magician who “explains but doesn’t show” than one who tries “to convince us about every little thing”. Because it’s quite human to be like “oh, yea that magic trick isn’t cool anymore” once you understand it. And His goal is to invite us to believe His son so that we’ll follow Him. His goal isn’t for us to know everything, though he is willing to teach us a lot. Remember: “ask and ye shall receive”.\n\nAnd now come, saith the Lord, by the Spirit…and let us reason together, ==that ye may understand;== D&C 50:10\n\nHe just didn’t say when He’d tell us.\n\n\nWhy he explains but doesn’t try to convince\nWhy do I believe that He likes to explain but not convince? A few ideas:\n\nHe taught people they would get to heaven by eating His flesh (sacrament), and people misunderstood and thought this was a “hard” teaching. They walked no more with him (John 6), but he didn’t run them down to try to bring them back. Why didn’t he?4\nHe was chained by the Romans, spat on, crowned with thorns, and whipped. But instead of showing them signs and wonders He let them abuse Him. Then He forgave them after they nailed Him to the cross multiple times and speared Him in His side and gave Him vinegar to drink (Psalm 69, John 19).\nHe knew He couldn’t convince them. He once healed a withered man’s hands and the Pharisees wanted to destroy Him because he did so on the sabbath (Matthew 12:9-14)5. How too are we like the Pharisees where we say “God shouldn’t be like X!!!” Shouldn’t we instead let God reveal Himself?\n\n4 Consider this in a humorous way: Why didn’t He say “oh sorry!! Please come back! I’ve got a quota of disciples to reach and if I don’t have you with me…””5 I love how instead of engaging with the Pharisees, He “withdrew” from them and “healed them all”, kind of as a divine rebel just saying, “you’re mad about me healing one guy on the Sabbath? Fine, I’ll heal everyone.”These three clues, among many others, give me perspective on why He’s okay being misunderstood by us.\n\n\nA God of purpose, children of purpose and promise\nWhy does He want us to live by faith? Lots of reasons. But there’s only one I’ll speak to: He’s a God of purpose:\n\n37 And the Lord God spake unto Moses, saying: The heavens, they are many, and they cannot be numbered unto man; but they are numbered unto me, for they are mine. 38 And as one earth shall pass away, and the heavens thereof even so shall another come; and there is no end to my works, neither to my words. 39 For behold, ==this is my work and my glory—to bring to pass the immortality and eternal life of man==. - Moses 1:37-39\n\nHis goal is to prepare an environment to teach and challenge us. Our goal should be to learn who, how and why.\nOur purpose should not be to put God in a box or imagine Him black and white. It should be to learn who He is, so He’a no longer an unknown God.\n\nAnd this is life eternal, that they might know thee the only true God, and Jesus Christ, whom thou hast sent. - John 17:3\n\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2023-12-16-humanity-and-ai/index.html",
- "href": "posts/2023-12-16-humanity-and-ai/index.html",
- "title": "What is humanity",
+ "objectID": "christianity/2023-12-17-christmas-celebrates-a-living-christ/index.html",
+ "href": "christianity/2023-12-17-christmas-celebrates-a-living-christ/index.html",
+ "title": "Christmas celebrates a Living Christ",
"section": "",
- "text": "AI and humanity\n\n\nI code in cursor. I use ChatGPT when that fails. I see Pika and Runway for text to video. I see midjourney for photos. Other things that just use GPT to make blog posts.\nWill AGI even solve the problem of getting everything to connect to everything?\nDo we want a stochastic assistant to have access to my bank account, text messages, medical records, social media accounts, etc?\nWill LLMs be the space between human communication? If we’re all using bots to generate material to send to each other, and all use bots to summarize that, then where will the real human connection exist?\nWe all want a world where our problems go away, but if we don’t have problems then are we human? To be naturally human is to be discontent and desire more, and it takes the wise to overcome these impulses. But the masses are not and never have been wise. Only the individuals.\nThe future is bright. But where does humanity fit into all of it?\nWhen problems are too easy to solve with AI, humanity gets to focus even more on the really hard problems: agreeing on how to get along.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Away in a manger.\n\n\n\nOld news\nElder Neil Anderson’s invitation here is worth listening to. It’s an invitation to ponder the reality of the birth of our Savior. It’s rather simple, but it evokes a profound question we should all consider.\n\n\nHow wonderful it is to celebrate the birth of our Savior and Redeemer. Jesus Christ is the central figure of all human history.I know He lives. I know He was born. I know He has been resurrected. I know He is our Savior and Redeemer. And with you I rejoice in this beautiful… pic.twitter.com/YDQpIzgk0n\n\n— Neil L. Andersen (@AndersenNeilL) December 17, 2023\n\n\n\n\nDid He come?\nNothing Elder Anderson shares here is new. It’s rather repetitive. Same old story. Same-old old-guy telling me to believe in a Jesus.\nIf you consider the most currently interesting things in this world, Artificial Intelligence, some may tell you that there exists a future utopia where we’re supported by infinitely intelligent digital agents who can solve complex world problems and bring us to…utopia.\nOk. We’ve also heard this before. The Tower of Babel tried to approximate heaven, and now humanity is trying to approximate Godlike intelligence.\nThe void of Christ in one’s life produces a desire to create Christ. A savior. Something to literally save us from our woes.\n\n\ni think AI is going to be the greatest force for economic empowerment and a lot of people getting rich we have ever seen\n\n— Sam Altman (@sama) February 13, 2023\n\n\nLet’s be real. I’m a tech optimist. I say bring it on. But I’m also a Christian. In my view, the most important question isn’t “what will AGI do to/for humanity?” Instead, the most important question is “was Jesus Christ the promised messiah?”\nIf Christ came, and was indeed the prophesied Christ, then all of life’s ultimate questions are already answered.\n\nWhy are we here? To die and be resurrected to live again.\nWhere did we come from? The same place He came from: the presence of God.\nWho are we? We are sons and daughters of God worth saving.\n\nYes, the questions are already answered: it doesn’t matter what happens with AI. Because the end is known: the Lord will come again.\nLet us treasure this time to reflect on a Savior who already came. Who lives. Who loves us beyond mortal (or AI) comprehension.\n\n\nWhat can we do?\nLet’s take a moment, perhaps an hour, to consider the miracle of Christmas:\n\n\nAs Christmas approaches, I realize that some may have concerns and perhaps some anxiety about the future. There may be a lot of “noise” in your life: more or less constant engagement online without downtime, without time to be quiet and reflect and think, without time to look… pic.twitter.com/4rqS2F6No9\n\n— D. Todd Christofferson (@ChristoffDTodd) December 15, 2023\n\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2023-12-24-twas-the-night-before-christmas/index.html",
- "href": "posts/2023-12-24-twas-the-night-before-christmas/index.html",
- "title": "Twas the Night before Christmas",
+ "objectID": "christianity/2023-12-24-he-is-the-light-that-we-should-share/index.html",
+ "href": "christianity/2023-12-24-he-is-the-light-that-we-should-share/index.html",
+ "title": "His is the light that we should share",
"section": "",
- "text": "Link to the 1912 publication, Twas the Night Before Christmas\noriginal poem was 1820’s\nSt Nick was 4th century\nDutch brought the name “Sinterklass” to the U.S. in the 1600’s.\nSt Nick becomes red by 1931, thanks to Coke\n\n\n\n\n\n\n\nThe OG St. Nick\n\n\nEverything below is produced by GPT:\n“How long has the concept of Santa been around? Where did”st nick” come from?”\n\nThe concept of Santa Claus has evolved over many centuries, originating from the historical figure of Saint Nicholas. Saint Nicholas was a 4th-century Christian bishop in Myra, which is now part of modern-day Turkey. He was known for his generosity and for performing miracles, which earned him a reputation as a protector of children and sailors.\n\n\nThe name “Santa Claus” evolved from the Dutch “Sinterklaas,” a short form of Sint Nikolaas (Saint Nicholas). Dutch immigrants brought the traditions of Sinterklaas to America in the 17th and 18th centuries. In these traditions, Sinterklaas was depicted as a serious, religious figure who rewarded good children and punished the naughty ones.\n\n\nDuring the 19th century, especially in America, the image of Santa Claus as we know it today began to take shape. The 1823 poem “A Visit from St. Nicholas” (commonly known as “Twas the Night Before Christmas”) played a crucial role in shaping the modern character of Santa Claus, including his physical appearance and the tradition of delivering gifts to children.\n\n\nLater, in the 19th and 20th centuries, Santa’s image was further popularized through various media, including advertisements, films, and songs, which contributed to the jovial, red-suited, and reindeer-driving character recognized globally today. This character represents a blend of the historical Saint Nicholas, folk traditions from various cultures, and modern commercial and media influences.\n\nHow did St Nick become red?\n\nThe transformation of Saint Nicholas into the red-suited Santa Claus is a gradual evolution influenced by cultural, artistic, and commercial factors. The image of Santa Claus wearing red clothes began to become popular in the 19th century, but it was solidified in the 20th century, particularly through advertising.\n\n\n\n19th Century Literature and Art: In the 19th century, Santa Claus was often depicted in various colors of robes, including blue, green, and brown. However, by the latter half of the 19th century, the image of Santa in a red suit began to appear more frequently in literature and illustrations. Thomas Nast, a prominent American cartoonist, played a significant role in popularizing this image. In his illustrations for Harper’s Weekly starting in the 1860s, Nast depicted Santa Claus in a variety of outfits, including the now-familiar red suit.\n\n\n\n\nCoca-Cola Advertising in the 20th Century: The standardized image of Santa Claus in a red suit with white fur trim was popularized and became iconic largely due to Coca-Cola advertising campaigns. In 1931, Coca-Cola commissioned illustrator Haddon Sundblom to create a Santa Claus for their Christmas advertising campaign. Sundblom’s Santa was jolly, rotund, and dressed in a red suit with white fur trim, which matched the company’s color scheme. These advertisements were widely circulated and became immensely popular, cementing the image of Santa Claus in a red suit in the public’s imagination.\n\n\n\nWhile the red-suited Santa Claus is now ubiquitous, it’s important to note that the evolution of his image was influenced by a blend of historical, cultural, and commercial factors over several decades.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "When Christ was teaching his disciples, he invited them to share their light to the world. But then he elaborated in an interesting way:\n\n3rd Nephi 18:24 Therefore, hold up your light that it may shine unto the world. Behold I am the light which ye shall hold up—that which ye have seen me do. Behold ye see that I have prayed unto the Father, and ye all have witnessed.\n\nThe Lord is their light. He is our light. He is the light we hold up to the world and let shine. He is our light that we let shine.\nThe pronouns here are impactful by their implication:\n\nHe assumes we have light\nHe assumes we have Him as our light\nHe invites us to share the light we possess\nThe light we possess is His light\nBut it is ours. And it is his. Our light is His light. His light is our light.\n\n“Let your light so shine before men that they may see your good works…”. Read in this new way: obtain the Lord as your light. Then shine His light before men.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2022-07-20-de-sciencing-data-science-and-talking-like-a-normal-person/index.html",
- "href": "posts/2022-07-20-de-sciencing-data-science-and-talking-like-a-normal-person/index.html",
- "title": "De-sciencing Data Science and Talking Like a Normal Person",
+ "objectID": "books.html",
+ "href": "books.html",
+ "title": "Book Reviews/Notes",
"section": "",
- "text": "Building a Data Science culture\n\n\n\n\n\n\nTL;DR: Always keep it simple and always bring it back to the business need.\n\n\n\nA coworker once asked me for advice on how to bring technical rigor into a non-technical culture.\n\n\nWhen the Simple Solution Wins\nA few years back I wanted to see if I could predict which NBA team would win in their next match. FiveThirtyEight is a blog that does this every day. Could I build a model as good as theirs?\nI built a few models and put up a site. Below I have a table of accuracies for each of my models compared to FoveThirtyEight’s.\n\nWhat I learned is that Nate Silver’s FiveThirtyEight three models are only 66% accurate. In 20 hours I got a model as accurate (V02 above) as his using just team level data (win %, recent win streak, etc.) and no player data (injuries, etc.). Also my model was built using data from 2005-2017 to predict 2019 performance. His model was updated daily with recent stats.\nBut even more glaring is that if I made a prediction on a single feature (Home % > Away %) I could achieve 63.7% accuracy. This feature is a 1 if the home teams win % is greater than the away teams win %, 0 otherwise. Meaning, the home team wins 63.7% of the time their season win % is higher than the away teams season win %.\nWas it worth building an XGBoost model with 30 features to get 65.8% accuracy when just a simple calculation is 63.7% accurate? Depends on business context and how valuable that 2% gain is and what the development/engineering cost of deploying a model is. But if I were you I’d start with the simple model to get it into production to start generating value and then come back later to optimize.\nCaveat: it took a lot of work to discover that feature. 90% of the problem was building the data set. So at the end, building a model was pretty simple once I had the data. (This is where cloud AutoML will take over going forward, sorry!.) And I needed the model to see the feature importance (this was the top feature.) This is a valid data science approach: use modeling to discover the patterns. But just because you have a model doesn’t mean it’s worth deploying.\nKeep it business focused: when building a recommendation system, start with “what’s most popular”. Why would you need K-means or whatever to begin? And can’t you build your own K-means by creating heuristic filters? For example: if you want to recommend me a shirt based on shirts I’ve searched, you can find all shirts of same color, price, and size and recommend those. Building a model to take into account dimensions, patterns, what others buy, etc is fine but should be a V2.\n\n\nMindsets\nThere’s a temptation to act like an academic professor and use fancy jargon when working with your coworkers. Even worse, there’s a temptation to think you’re smarter than your coworkers because you know this mathy thing that they don’t. But remember this quote:\n\nWhen they are learned they think they are wise…1\n1 2 Nephi 9:28\nThis pride power struggle is your downfall for two reasons:\n\nIt blinds you to thinking your solution is good just because someone else can’t understand it.\nIt blinds you from focusing on what matters most: getting the simplest and best solution.\n\nYour coworkers aren’t your students eating you. They’re your smart coworkers solving the same problems you are.\nA better mindset is that of a professional athlete, like a football player. They might be dumb as bricks at times but they’re all incredibly smart players. Can you explain your technical solution like a pro football player would?\n\nBad: “I built an XGBoost classification model that reduces the RMSE by minimizing the Gini coefficient on each node…”\n\nBetter: “I noticed that most users close the app after they reach this page. I tried looking at some reasons why and found a few common behaviors (DISCUSS, get feedback). I think we can predict when this will happen and here’s a framework to do that.”\n\nThere’s a temptation to think “I did all this technical work, I should get credit for it!” My manager once coached me, “Your stakeholders should be amazed by how simple you made it. Not by how complicated the problem is.”\nThe best example I know of this is Chris Voss, an FBI negotiator. Listen to his podcasts or YouTubes. He’s a humble guy and doesn’t get into the technical stuff. He knows the principles and keeps the explanations simple. He helps you feel smart when you listen to him. He’s one of the best negotiators in the world and if you were having dinner with him you’d think he was a normal guy. This is how data scientists should be.\n\n\nBugs\nThere are two types of bugs in data science: Software bugs and data bugs. You’ll spend 80% of your ramp up time on a project setting up your environment, learning the data, trying to understand the primary keys and logging and unique values, etc.\n\n\n\n\n\n\nNobody cares about how you do any of this (the same way you don’t care about how cereal ends up in your grocery store), but they do care that you do it because it shows that you’re thinking about the right things.\nThey’ll also worry that you’re spending too much time on the wrong thing, so make sure you believe your actions are solving a business purpose.\n\n\n\n\n\nTakeaways\n\nFocus on principles, not technical details.\nGet the simplest, useful solution into production first to start creating value. Then figure out if it’s worth optimizing. If your product gets deprecated in 3 months then the fancy model will be wasted time anyway.\nYour stakeholders should be amazed by how simple you made it. Not by how complicated the problem is.\nIf you can’t explain a trend with a simple bar/line chart, a fancy regression model won’t do.\nStart with the non-technical (basic math) solution. Then see if you can beat the basic math with a fancy model. Evaluate the cost/benefit of the technical solution relative to the dumb version.\nwhen you have a fancy solution, make sure you compare it to the basic solution. Make sure you can beat the solution your teammates come up with. And make sure you carefully lead them from “their” solution to yours. “At first I tried X. X was fine. But X didn’t account for Y. So I built a model M and saw that M out performs X by 10%.”\nJust because you know/use math, doesn’t mean it’s right for the business. Data Scientists can easily be myopic because the work we do is challenging and deep in the weeds.\nYour boss doesn’t care about the nature of the bugs, but you should try to bring first principles to the table so she can cover for you. “I keep hitting this bug in the data. Can we ask/work with Infra team to solve this because it eats my time.”\n\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Goal: To spend less time consuming more, more time consuming intentionally.\nMostly incoherent bullet points of thought provoking moments.\n\n\n\n\n\n\n\n\n\n\n\n\n101 Essays that will Change the Way You Think, by Brianna Wiest\n\n\nWe all started as strangers. It’s a choice who remains one.\n\n\n\n\n\n\nDec 19, 2023\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nDune\n\n\nThe plans and risks of generations\n\n\n\n\n\n\nDec 23, 2023\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nGetting Things Done\n\n\nA powerful system to be more present.\n\n\n\n\n\n\nJun 24, 2022\n\n\n\n\n\n\n\n\n\n\n\n\nHitchhiker’s Guide to the Galaxy\n\n\nHow did it take me so long to finally read this?\n\n\n\n\n\n\nDec 18, 2023\n\n\n\n\n\n\n\n\n\n\n\n\nScientific Freedom\n\n\nInnovation comes by varied incentives: intellectual curiosity and economic incentive\n\n\n\n\n\n\nDec 12, 2023\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nThe Best of Edgar Allen Poe\n\n\nYou think me mad?\n\n\n\n\n\n\nDec 23, 2023\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nThe Killers of the Flower Moon\n\n\n\n\n\n\n\n\n\nDec 22, 2023\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nThe Little Red Hen\n\n\nPeople love to enjoy the fruits, but who will bake the bread?\n\n\n\n\n\n\nDec 20, 2023\n\n\n\n\n\n\n\n\nNo matching items\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only.",
+ "crumbs": [
+ "Book Reviews/Notes"
+ ]
},
{
- "objectID": "posts/2022-06-25-how-i-use-ticktick/index.html",
- "href": "posts/2022-06-25-how-i-use-ticktick/index.html",
- "title": "How I use TickTick",
+ "objectID": "book-reviews/2023-12-20-the-little-red-hen/index.html",
+ "href": "book-reviews/2023-12-20-the-little-red-hen/index.html",
+ "title": "The Little Red Hen",
"section": "",
- "text": "I’ve recently read “Atomic Habits”, “Deep Work”, “The Power of Habit”, “Getting Things Done”, “Essentialism” and “Effortless”. These productivity paradigms influence this post.\nInspired by this Reddit post, I’ve tried thinking how I want apply the principles of GTD and these other paradigms. I think GTD is great for me because it helps me get ideas out of my head and Essentialism is a better pursuit to make sure I’m not caught up in minutia. I don’t see them as incongruent, but one as an ideation organization approach and the other as a relentless focus.\nFive steps of GTD:\n\nCapture\nClarify\nOrganize\nReview\nEngage\n\n\nCapture\nI have an iOS shortcut that lets me quickly add a TickTick task to my inbox. I put this on my Home Screen for instant access and to avoid distraction. I want to capture the idea, not open TickTick and get distracted by other ideas.\nOr I use the TickTick today widget and upload it to today directly. Depends on the task and how detailed I want to be.\n\n\nClarify\nHere’s the system I use to clarify.\nDefinitions:\n\nTask: an action item.\nProject: an intended outcome with multiple tasks.\nEpic: an intended outcome with multiple projects.\nKey Result: high level, long term accomplishment. (Increase revenue 10%)\nObjective: a squishy, feel-good purpose statement. “Deepen relationships with loved ones.”\n\nTask grooming process: Groom the inbox every night and identify the intended outcome and next action. Tag the task.\n\nIf the task was “send Birthday card”, that’s really an intended outcome and is a small project. The tasks are 1) buy card from CVS, 2) put in envelope with stamp, 3) drop in mailbox. These three subtasks need to be understood. But a separate inbox entry may be “pay phone bill by sitting down at my computer at 10pm”. This isn’t a project. It’s just a one off thing. But it can still be clarified with details.\nLabel if it’s a project or not. Use iOS shortcuts to make this really easy. For example, gg = “🎯 #Project”\nThe challenge is if I should have things as projects or as individual tasks. I would say that as much as things can be independent, keep them tasks. Like “apply for credit card1” and “apply for credit card 2” are both personal finance tasks. But they are independent.\nA task isn’t necessarily short and a project isn’t necessarily long. It’s just whether this Indy ended outcome has multiple steps or not that might be done over a few days, whereas a task is probably done in one sitting.\nIt’s assumed that unless something is a task, it’s a project.\nEvery epic has its own name. I nest epics under “Work > #Epic1, #Epic2”, etc. This helps me organize my work deliverables. I have like 5 epics right now at work that will span several quarters.\n\nThen I may choose to add tags with metadata, which I haven’t figured out if this is useful yet. But “Atomic Habits” says the best way to get stuff done is to stack it. So here is my tagging system to help me stack:\n\nCommitment: ASAP, Eventually, Someday Maybe. Will I do this task eventually? Pay a bill - yes. Learn a language - someday maybe.\n\nDuration: 5min, 30min, 1hr, 2hr, 4hr. 9hr (all day)\nLocation (where to do the task): Home, Desk, Car, Backyard. Most work things are done at my desk. But perhaps some can be done while taking a walk. Most personal things are done around the house or on the couch at night.\nTime of Day: “1: Pre work”. “2: start of workday. 3: lunchtime, 4: afternoon, 5: after work, 6: nighttime”\nFocus: “Deepwork” or “shallow”. Deep work should be uninterrupted and be 3 hours long.\nSeverity: S1: if I don’t do this nothing will happen. S3: not doing this will have bad consequences or lots of regret. S2 - in between :).\nDeadline: I’m still thinking about this one. To me there’s a difference between mowing the law which has a linear cost of looking worse each day for not being cut and failing to pay my taxes, which has festeringly worse consequences the longer I put it off (pay a fine, end up in jail, etc.). I guess this is captured implicitly in “ASAP” commitment. But paying my taxes is only a cost (preventing negative outcomes) versus the ASAP of getting my project finished (positive consequences).\n\nNone (default): no consequence for never doing this.\nBinary: You either make the deadline or never and this opportunity ceases forever.\nRolling: If you miss the deadline there will be another.\nFestering: the longer you put this off the worse it becomes, but no deadline.\n\n\nAs I write this, this sounds like overkill…classic me. I sincerely doubt it’s useful to of all of this. But having the system available will let me use what I need when I need it. If I’m still using any of this in a month that’ll be a miracle. But I’m going to start with something in mind and go from there.\n\n\nOrganize\n\nOnce clarified, move out of inbox into a list “work” or “personal”.1\nPrioritize.\nDelegate. (Myself, someone else.)\nAssign due dates and times (if not already done)\n\n1 Tags and lists are identical except Lists have sections and views (such as Kanban). In a list you can sort by tag and vice versa. But each task can only appear in one list whereas tags are infinite. I noticed that choosing which list something belongs to can be cumbersome if it’s in a nested list, so I want only high level lists (no folders) and have this be simple. I also have “personal” and “work” tags which have projects within them, so I don’t really see why I’d need to duplicate this. On second thought, each task can only belong to one work project so it’d make sense for this to be my grouping. I suppose I just don’t want to manage all that. Or, I chose to use tags to manage my projects because I was cheap and I can only have 9 lists on the free version but unlimited tags? Either way this is easy to reorganize later if I choose to switch. Tags it is for now.\n\nReview\n\nEach night I will review the days tasks and re-prioritize as needed.\nPin the one task/project I want to get done that day.\n\nEssentialism says to take the 6 tasks of the day and choose just one. Then identify what else you need to say no to to ensure you have the energy to accomplish that task.\n\nWeekly planning session to review all of this.\n\n\n\nEngage\n\nDon’t work on any tasks in a day/week that weren’t planned for that day (unless something comes up).\n\n\n\nFinal Thoughts\nA task system should be saving me time, not consuming it. It should be relaxing me and my anxieties, not amplifying them. I’m going to give this 30 days and see if my stress decreases and productivity increase. If so, I’ll keep using this system.\nGTD states that by not writing down all the ideas, you waste energy because things come back to nag you later. But surely energy is wasted by organizing a valueless idea.\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "TL;DR: Be the hen. Do work!\n\n\n\n\n\n\n“The Little Red Hen” is a classic folk tale that tells the story of a hardworking red hen who finds wheat seeds and decides to plant them. Throughout the story, she asks other farm animals for help with various tasks—planting the wheat, harvesting it, milling the flour, and baking the bread. Each time, the animals refuse to help. However, once the bread is baked and the hen asks who will help her eat it, all the animals eagerly volunteer. She declines their offers, choosing to eat the bread with her chicks, who assisted her in the work.\nSo, too, do startups everywhere struggle to get funding. 100 rejections, etc.\nAnd great products don’t get attention until there’s social proof. Many e-commerce companies understand this - sales take off as more people buy, which gets more reviews, which gets more trust - even if the product didn’t change much between first sale to the last.\nThe only difference between startups and the red hen is she perhaps had a recipe. She didn’t have to wonder, “is baking bread possible? Will the recipe be delicious?”\nIt takes a founder’s belief to see what isn’t yet so and build it anyway. Be the founder. Be the hen. Dream big. Go bake some bread even if nobody’s watching.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2023-12-15-llms-can-find-a-needle-in-the-haystack/index.html",
- "href": "posts/2023-12-15-llms-can-find-a-needle-in-the-haystack/index.html",
- "title": "LLMs can find a needle in the haystack",
+ "objectID": "book-reviews/2023-12-22-the-killers-of-the-flower-moon/index.html",
+ "href": "book-reviews/2023-12-22-the-killers-of-the-flower-moon/index.html",
+ "title": "The Killers of the Flower Moon",
"section": "",
- "text": "Is RAG necessary when you have incredible memory?\n\n\n\n\n\n\n\nContext\nCheck out this thread:\n\n\n(1/8) The Needle in the Haystack done by @GregKamradt was an amazing analysis of retrieval performance! Greg has graciously allowed us to build on his work with a repository that is now OSS.@natfriedman We have a much more rigorous test we’ve put out based on this idea.… pic.twitter.com/i5O8zrcwQT\n\n— Aparna Dhinakaran (@aparnadhinak) December 15, 2023\n\n\nThis is a powerful analysis. Sure, Anthropic will find a way to improve or challenge the results. But the point is clear: these technologies can remember hyper specific 7-digit random numbers out of a batch of 126,000 tokens, where a token is roughly 4 characters. GPT is clear winner here, too.\nAlso, open source is getting incredibly good. This implies the future is open source.\n\n\nComparing @OpenAI #GPT4 Turbo to @MistralAI GPT-4 is pretty good in that region in general. Interesting to see how @MistralAI scales to larger context windows pic.twitter.com/WQo6MmGIHh\n\n— Aparna Dhinakaran (@aparnadhinak) December 15, 2023\n\n\n\n\nImpact\nRAG can be used to make retrieval more efficient. But if retrieval is already super efficient maybe RAG is only a short term thing. Context lengths of 10m tokens…probably by next year right?\nStart of the year we were at 4K tokens. Now there are 126,000 tokens. 30x improvement. So to do another 30x improvement is 3.76M. So yea, by next year you should be able to just load the entire RAG database into memory. But…gonna be super expensive.\nPoint is: would GPT be this effective if it was using RAG over a database? Or is it more effective loading it all into context?\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "TL;DR:\n\n\n\n\n\n\nOn Spotify\n\nHale: self made man. Cowboy. Learned the trade. Got one cow. Then many. Then had 25,000 acres.\nAlly. Used his position as a white guy to advocate on behalf of the Osage.\n\nCherokee nation parceled and “first come first serve” over 42,000 acres\nWild West: private eye 1921 > than local sheriff\n1850 Alan Pinkerton - private eye.\nguardians - Osage couldn’t spend their own money without approval\ncriminal’s secret eats away at him\njudicial execution is a challenge of taming the evils within even those on the “right side of the law” calling for justice. Do we really need to execute someone? What if they’re actually innocent?\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "gratitude/2023-12-25-grateful-for-doctors/index.html",
- "href": "gratitude/2023-12-25-grateful-for-doctors/index.html",
- "title": "I’m grateful for doctors",
+ "objectID": "book-reviews/2023-12-23-the-best-of-edgar-allen-poe/index.html",
+ "href": "book-reviews/2023-12-23-the-best-of-edgar-allen-poe/index.html",
+ "title": "The Best of Edgar Allen Poe",
"section": "",
- "text": "The Healer’s Art\nI didn’t break my first bone until I was 17, but I visited my first doctor many years before.\nI grew up with an odd fever syndrome that came and went every six weeks. 105 degree fevers each time. I remember two distinct moments hallucinating that the room was splitting and shifting and another when bees were covering the corner of my bunk. The delirium was real.\nEventually it went away. But there was also a doctor there each time I got sick. One for me, to take my blood. One for my mom, to try to comfort and counsel her.\nI’m grateful for doctors of the mind who help alleviate sicknesses of stress, anxiety, depression, etc. I know more people with some type of recurring emotional distress than I know without. I see the benefits reaped by those who seek emotional healing through therapy. I’m proud of the friends who struggle, but who don’t struggle alone. The only way out, is with.\nI’m grateful for healers of the soul. Christ invited everyone to change their ways by letting go of the things that didn’t bring them true happiness, and focus on the things that do. In one invitation, he urged us to come unto him “that I may heal you”. I’ve witnessed spiritual healing through prayer, which is the process of going from feeling forgotten to then knowing I am a child of God and he has a plan for me.\n\n\nHe Came to Heal\nThis to me is the meaning of Christmas: healing. Freedom from bruised bodies, broken minds, or hurting hearts. He came to be broken, so that we may be healed.\n\nBut he was wounded for our transgressions, he was bruised for our iniquities: the chastisement of our peace was upon him; and with his stripes we are healed. Isaiah 53:5\n\n\nHe will take upon him their infirmities, that his bowels may be filled with mercy, according to the flesh, that he may know according to the flesh how to succor his people according to their infirmities. Alma 7:12\n\nNot all healing happens in an instant, and sometimes some things don’t fully heal or heal at all. But I’m grateful healing is possible and there are doctors in the constant pursuit of it. And I’m grateful for friends who have, at times, helped heal me in all these ways.\nMost of us won’t be known as doctors. But all of us try the best we can to help heal our loved ones, such as healing a child with a band aid, a spouse with a hug, a friend with a lunch. Thank you for doing the work of Christmas, all year round.\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Raw Notes\nThe Tell-Tale Heart:\n\nyou think me mad. Nah, I’m cool. But the man grows increasingly mad.\nguilt follows the commission, before then the guilt is only in the hesrt. After it consumes.\n\nThe Cask of Amontillado:\n\nwatch your pride. Don’t let it trick you into walking into catacombs\n\nThe Masque of the Red Death:\n\nfruitless to avoid it\nallegory of phases of life\nPrince Prospero tries to fight death, but dies in the process\n\nThe Raven\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "gratitude/2023-12-24-light/index.html",
- "href": "gratitude/2023-12-24-light/index.html",
- "title": "Light",
+ "objectID": "book-reviews/2023-12-12-scientific-freedom/index.html",
+ "href": "book-reviews/2023-12-12-scientific-freedom/index.html",
+ "title": "Scientific Freedom",
"section": "",
- "text": "I installed 72 lights today. This took drilling a hole into my fence, then making the cord taught, then screwing the screw into the cord’s loopy thingy. Then I move on to the next one.\nThe whole process took three hours. But I’m glad it took three hours. I appreciate it so much more because I did it myself. I thought carefully about each hole I drilled.\nAnd it felt good to sit out there and listen to an audio book (lots of Dune today, some Killers of the Flower Moon).\nI’m grateful for Thomas Edison. For the inspiration he received, for the hard work he did to give the gift of light to the world. And if he hadn’t done it someone else would have, which is also something to be grateful for. There are a lot of smart, good people on this earth.\n\n\n\nNot what it looks like, but how it feels\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Max Plank, GPT 4\n\n\n\n“Scientific Freedom”\n\n\n\n\n\n\nTL;DR: a scientist isn’t a title, but a philosophy\n\n\n\n\n\n\nScientists are hampered by the peer review process. The notion that some tolerate the process so they can get their grants approved and get tenure, and others eventually accept the process because that’s how they got tenure, is a sad outcome of human incentives.\nInterestingly, inventions have always happened regardless of peer review. Edison didn’t need someone to review his invention to tell him whether or not it’s a valid invention. Either the light bulb illuminates or it doesn’t.\nYet that’s exactly how things go now. So what’s the result?\nPeople write to please the peer reviewer. Social conformity. Group think.\nPeople are incentivized to game the system: write a grant that gets them funding so they can get tenure and support their families.\nThe incentive to be innovative is small, though intellectual curiosity powers through for some. The incentive to stay in line and do what’s necessary to get approved is strong.\nWe’ve also learned as a consequence of 2020-2022 that a lot of scientists can indeed be wrong. Today science is becoming less about discovering truth and more about people owning their truth…people love their politics.\nWhat makes a scientist anyway? Am I a scientist because I’m a data scientist? Because I studied statistics? I’ve never published an academic article in a journal, so is that the threshold?\nI think someone who pursues truth through repeatable experimentation is a scientist. Back in Plato’s day they were just called philosophers.\n\n\nToday’s incentives\nThere’s a reason really smart people now work at Apple to invent the iPhone, Facebook to work on LLMs, and SpaceX to build rockets. The freedom they feel at these companies is empowering. Per “Scientific Freedom”, if the trends continue then people like Max Plank would never have been able to get the funding needed to invent what he did. Where’s the money? Social media. Advertising. And now, AI.\nPeople will always be inventive. But it’s by removing the social barriers that they become so.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "lists.html",
- "href": "lists.html",
- "title": "Lists of Things",
+ "objectID": "book-reviews/2023-12-19-101-essays-to-change-the-way-you-think-by-briana-weist/index.html",
+ "href": "book-reviews/2023-12-19-101-essays-to-change-the-way-you-think-by-briana-weist/index.html",
+ "title": "101 Essays that will Change the Way You Think, by Brianna Wiest",
"section": "",
- "text": "Ongoing bookmarks and lists of things I want to remember.\n\n\n\n\n\n\n\n\n\n\n\n\nCool AI Tweets\n\n\nRunning list of cool things I’ve seen.\n\n\n\n\n\n\nDec 14, 2023\n\n\n\n\n\n\n\n\nNo matching items\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only.",
- "crumbs": [
- "Lists of Things"
- ]
+ "text": "TL;DR: 101 ideas. This is like a modern day Meditations.\n\n\n\n\n\n\n2023-12-20 9-11\n\nhappiness of pleasure, grace (appreciation), excellence (emotional resilience)\nsome of the poorest countries are the happiest.\n\n2023-12-19 Essays 1-8\n\nwe all started as strangers\nto change your beliefs, go have experiences that manifest and confirm the beliefs you want to have\ncan only have a few truly\nour relationship with others is a pure reflection of our relationship with our self. Work on your relationship with yourself to improve your relationship with others.\ndo you look at your leg and say “that’s me?” No. That’s my leg. Do you look at your eye, heart, hair? No. Do you look at your fears, hopes or dreams and say, that’s me?\nfear and Anxiety aren’t feelings. They’re symptoms.\nWhat emotion do you feel the most often? Is it positive? If not, why does a negative feeling deserve the spotlight? What does this tell you about your relationship with yourself?\n\n2023-12-22 essays 12+\n\nknowing doing gap. “If I lived today on repeat for 10 years where will I be?” What would happen if I didn’t do X?\n\nInsights: - self actualization is what we’re all doing. We all mostly have shelter, food, job, even if minimum wage.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "christianity/2023-12-24-the-christ-child/index.html",
- "href": "christianity/2023-12-24-the-christ-child/index.html",
- "title": "The Christ Child",
+ "objectID": "book-reviews/2023-12-23-dune/index.html",
+ "href": "book-reviews/2023-12-23-dune/index.html",
+ "title": "Dune",
"section": "",
- "text": "The story of the Savior’s birth brings many wonderful concepts to memory and can teach us a lot about what will bring us true joy and happiness in this life.\n\nThis is my favorite video of the Christmas story. The most powerful part is the moment when Mary sings a lullaby to Jesus. I remember reading somewhere that this moment was completely impromptu by the actress, and it speaks volumes to the sacredness of the moment where Mary first laid eyes on her beloved son.\n\nSo what does this story give me perspective on at this time of year?\n\nChrist was the rightful, bloodline heir to King David’s throne. He was the King of the Jews. But he never sought authority. He never sought a throne. He instead taught “Render therefore unto Caesar the things which are Caesar’s”. He didn’t engage in turf battles. In rebellion. In insurrection. He just wanted people to find happiness in their circumstances by living with lovingkindness in their hearts for their God, their neighbor, and themselves.\nIn his role as a Savior and Judge of the world, he sought to not condemn the sinner trapped in the act of sin. Instead he came to liberate. To forgive. To support.\nIf you believe Christ was the Creator and Savior of the world, consider how He chose to come into the world. He was born in a stable, mucked with animal feces and rotting food scraps, covered with the stench of…a stable. He was rejected of a room in the inn. He was laid in a manger, perhaps a feeding trough. Maybe Mary had help giving birth, maybe she didn’t. He was born generally alone. I wonder if He chose to do his birth this way because nobody can say he had it easy. He didn’t have the special treatment of a King’s birth. No, he was born to an exhausted, weary traveler, likely on the cold hard floor of a dirty barn.\nIf we think about the Savior being the architect of this birth, think about how he planned for angels to show up to shepherds. Not King Herod. The shepherds. He would be the prophesied sacrificial Lamb. He would be the Good Shepherd. Yet these people were perhaps the minimum wage, forgotten ones outside the city gate. Yet these are those who had the privilege to see Him first. And He invited them to come and see. His whole life was an invitation.\nHe also invited the rich wise men. All are invited. He wanted all to see, and he wanted the world to know that He was for everyone regardless their circumstances. And that all of us, whether destitute or traveling with caravans, can bring the gifts of our heart to Him. We don’t need to bring gold. He just wants a “broken heart and contrite spirit.\n\nHow can we apply these things to our life, today?\n\nSeek out the forgotten. The people in our life that seem forgotten. The servants in our life (person on the other end of the customer support line, the checkout person, your supervisor) but even, your best friend who seems to be doing great…\nSeek not riches or wealth. Harod sought these things. Christ sought relationships and truth.\nDon’t seek for power. Christ deserved it. Christ had it. He turned water to wine, cursed a tree, and made a host of men fall over at the sound of his voice. Yet he didn’t touch the Romans.\nSeek to invite. To share. To uplift. Christ made shepherds feel important. He invited them to his crib side. And ” his hand is stretched out still.” (Isaiah 9:12)\n\nHe came like all of us, as a child. But he came to show us how a God would live. And he did.\n\n\n\nThe Christ as a Child, before worshipping Wise men\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "shield your son too much and he’ll not grow strong enough to fulfill any destiny\ndeprogram after the machines. Let the machines do too much thinking.\nhumans must never submit to animals\n“how did the woman seal my tongue?” Paul can’t tell his father he’s going to die. Why? Sometimes our tongues are bound to speak that which can’t be changed\n“the mind commands the body and it obeys. It commands itself and it rebels”\n\nConnections: - Alan Pinkerton killers “secrets break the character of the criminal till they give it out”. Dr Yueh cries when his scripture is given out.\nNotes: - if I made him do it it wouldn’t be his doing. - storytelling: why tell us 100% of everything up front? The movie hides the secrets about Yueh. - storytelling: the hindsight narrator\nWhen God has appointed a man to die, he guides his wants to desire that place\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "christianity/2023-12-13-three-little-pigs/index.html",
- "href": "christianity/2023-12-13-three-little-pigs/index.html",
- "title": "Three Little Pigs",
+ "objectID": "book-reviews/2023-12-18-hitchhiker's-guide-to-the-galaxy/index.html",
+ "href": "book-reviews/2023-12-18-hitchhiker's-guide-to-the-galaxy/index.html",
+ "title": "Hitchhiker’s Guide to the Galaxy",
"section": "",
- "text": "_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "The book of infinite possibilities\n\n\nThis is a classic. First time reading it and I love it.\nWhat I love about the storytelling:\n\nDeclare something as unavoidably true, even if ridiculous, and build all content off of it.\n\n\nNobody cares that the infinity probability drive is absurdly improbable. But once the reader just accepts this as fact the rest of the story make sense.\n\n\nget into the mind of the reader\n\n\nsperm whale? I love when he’s detailing the whales first and last thoughts. Just so random, but hilarious.\n\n\nbe spontaneous, but connected.\n\n\nI recently watched the second season of Loki and couldn’t stand it. It was all over the place. Random here, random there. But this is great because the spontaneity flows within the confines of the story.\n\n\nlogic, with social proof, can invert the mind\n\n\nArthur dent is flabbergasted by mice being the most intelligent species. Well the logic is sound: they were just fooling us and running experiments on us the whole time so we wouldn’t think anything of it. This flip on expectation, backed by “isn’t it obvious?!” gets the reader wondering for a second before they snap back to reality: no, mice aren’t running the universe.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "christianity/2023-12-13-three-little-pigs/index.html#first-glance",
- "href": "christianity/2023-12-13-three-little-pigs/index.html#first-glance",
- "title": "Three Little Pigs",
- "section": "first glance",
- "text": "first glance\nYes this is a book review about “The Three Little Pigs”.\nThe first pig is lazy. So he builds his house of straw.\nThe second is lazy too. So he builds his house of sticks.\nThen the third is the smart one who builds his house of bricks.\nBut the first two brothers just make fun of the mason. Why waste your time?!? We’re having fun!\nSo the wolf comes and blows down the houses of the lazy brothers. They run and hide in brick brother’s house.\nMoral of the story: be lazy but have a great, well prepared brother!"
+ "objectID": "book-reviews/2022-06-24-getting-things-done/index.html",
+ "href": "book-reviews/2022-06-24-getting-things-done/index.html",
+ "title": "Getting Things Done",
+ "section": "",
+ "text": "Principles and Core Insights\nAfter 21% my takeaways are:\n\n“My mind is weighed down by all of the ideas of things I can do.” I waste my time on todos that are just “go to doctor”, but should be “schedule appointment to go to doctor”. Anything that only takes 2 min should be done immediately. Anything longer later. But I love that clarity - I’ve often written lots of todos but not from the “next action to take” perspective.\n“Visions can be built from the ground up rather than just top down.” I’ve tried writing down big visions for my 10 year future and it never works for me. I’ve got anxiety when I try to think where I want to be in 10 years. I mean, I know I want to be happy, be best friends with my wife and kids and have meaningful relationships but whether I’m still a data scientist or a farmer by then is anyones guess. (I wanna keep my options open.) But as I look at the things on my todo list and “eventual” to do list, a different picture emerges. I know I want to be a writer (hence this blog), I know I want to travel to XYZ location, to be involved in giving back, etc. TLDR: I know things I want to experience in life. I don’t have to craft value statements and invent ideas of what to do (seems forced). Instead I can reflect on what’s currently exciting and important to me and categorize those things into high levels values/missions. This will help me when I need to prioritize so not every whim of an idea gets my attention.\nGTD book just lays out what a todo is. It breaks down the problem of doing things (big and small) into what they are (time sensitive, someday, etc). Whether I’ll follow the system is up for debate, but I love the ideas so far. It very much fits my naturally organized mind, but may not work for more spontaneous folks that don’t like as much structure and list keeping.\nMost of my “todos” have been “intended outcomes”, but not “next actions”. For example, I see I have “change car battery” on my list right now and “plan yosemite trip”. But these aren’t todos. I can’t organize and prioritize outcomes. I can only organize and prioritize actions, he says. Changing these to “call mechanic to book oil and battery change and identify an hour in your calendar”. And “open recreation.gov when walking around the house and find a campsite.”\nOrganize tasks around where you’ll do them. This is in line with Atomic habits which states to be hyper specific about where and when you’ll do something. “Call grandma when I get into my car tomorrow morning to go to work.” Or “take a moment at my desk during lunchtime to find a mechanic by opening Google Maps and plan an oil change. Find 1 open hour on the calendar.”\n\n\n\nPutting it into practice\nI’ve been using TickTick for my task manager, which has Lists (folders) where I can put my tasks and tags so I can label and categorize each task. Here’s my organization strategy:\n\nWhen jotting ideas write intended outcome and next action as my todo.\nTag with where I’ll do it and provide day or time it’s due (if applicable)\n\nMove each item to a respective list:\n\nASAP - there’s a useful urgency to this task (begin memo)\nEventually - there’s no urgency but I know I’ll want to do this\nSomeday maybe - no urgency and no commitment, but it’s an idea of something I want to do.\n\nTagging system: tag each item with either Work or Personal. Then within that, some meaningful themes:\n\nProjectName (work deliverable or personal project like “ImproveMyHealth”, “BuildApp”)1\nLocation (where I’ll do it). Around the house. At desk. In car. While cleaning the house, etc. this is habit stacking. Most work stuff is done at my desk.\nwhen I’ll do it. Stack with other activities. Examples: Before lunch.\nDuration (how long it takes: 1, 2, 4 hours, 5, 30min). This might be useful for planning.\n\nList system. Move from inbox to list when properly categorized. I like “Work” and “Personal” lists. I think since you can use unlimited tags, and tags can be nested in TickTick, there’s no substantive differ ne between list and tag. Not sure yet how to use the TickTick lists when tags are so useful. Lists of work and personal.\n\n1 Deciding Vision: As I write this, choosing a project might be a good place to insert a goal (intended outcome) as a project name. Instead of Exercise, the intended outcome is improving health. The goal is being able to run a 5k. So perhaps goal/objective is “ImproveHealth” and KR is “run 5k”. These goals can be identified by asking the 5 whys. If I have an idea if “go running” as my task, a good grounds-up vision exercise would be to ask “why do I want to run a mile tomorrow? Is it to accomplish something or to relax my mind? Why is relaxing my mind important to me? Why is my mental health important? Etc. I think subconsciously ideas come of what to do next all the time “I should work out” but rarely do I take the time to ask “why do I see the need to work out? What’s the greater vision here for this task?” Understanding that vision can help prioritize and build a desire to accomplish the task. If the idea “learn a new language” comes up, asking why can help me realize “oh, this is just for fun and won’t be for anything meatier than having fun! No pressure on this task and maybe I’ll never do it.” The nag to run a mile might help me realize “I need to take care of my health, and this can’t wait.”In summary:\n\nIdea/task comes.\nDocument in inbox.\nClarify the intended outcome and the next action.\nOrganize: tag. Set up reminder. Move out of inbox to list.\nReview and Prioritize: understand where these ideas will take you and choose your adventure.\nDo!\n\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "christianity/2023-12-13-three-little-pigs/index.html#not-quite",
- "href": "christianity/2023-12-13-three-little-pigs/index.html#not-quite",
- "title": "Three Little Pigs",
- "section": "not quite",
- "text": "not quite\nI’ve always been bothered by this as I read to my kids. But I’ve recently changed my opinion. Because im very much the first and second brother. We all are.\nWere inadequate. Maybe we’re not smart enough to know how to create bricks and such. Maybe we don’t have enough resources and the straw is the best we can do. And maybe we hide behind these insecurities by laughing at those who have more.\nMaybe we hide behind our insecurity and question God. The Wise One.\nSee, Christ is our Brick. He’s the safety. He’s the home we can run to when the world blows us over. He’s the Prepared One who was wishing for us to come to him the whole time. The cry of “you’ll be sorry” by this brother could perhaps be His loving warning of “please be wise. There’s a wolf out there.” Yet we don’t listen. We want to play.\nI love the third pig because he lets them in.\nAlso, the third pig never had to worry about almost getting eaten. Sure, first two enjoyed their fiddle and dance. But they almost died. The third pig is like the non-prodigal son. The one who stayed home and enjoyed the safety of the home the whole time.\nBe wise. Be forgiving. And yes, we always have a wise elder Brother who will accept us."
+ "objectID": "lists/ai/index.html",
+ "href": "lists/ai/index.html",
+ "title": "Cool AI Tweets",
+ "section": "",
+ "text": "TL;DR: AI is wild. Last updated per last headline. First created 2023-12-14\n\n\n\n\n\n\n\nv0.dev\n2023-12-15\nv0.dev a UI generator by vercel. React.\n\n\nCoffee by Coframe\n2023-12-15\nBuild React UIs super fast. GitHub\n\n\nAnnouncing Coffee: build and iterate on your UI 10x faster with AI ☕️👇https://t.co/0vBssgp0ue pic.twitter.com/JqwC8WpDzs\n\n— Coframe (@coframe_ai) December 13, 2023\n\n\n\n\nPrompt Engineering\nAdded 2023-12-15\nPrompt Engineering tips by OpenAI\n\n\nGenerate a Website from Screenshot\n2023-12-14\n\n\nscreenshot-to-code: upload a screenshot of any website, watch as AI progressively builds the html, iteratively improving the generated code by comparing it against the screenshot repeatedly. pic.twitter.com/TCrbJj3VS0\n\n— Siqi Chen (@blader) November 21, 2023\n\n\n\nGitHub\nhttps://screenshottocode.com/\n\n\n\ntldraw\n2023-12-14\n\n\nLooking to learn about the AI thing we made?👇Link in the next post pic.twitter.com/1FjY3eMSSF\n\n— tldraw (@tldraw) November 21, 2023\n\n\n\nhttps://makereal.tldraw.com/\nhttps://open.substack.com/pub/tldraw/p/make-real-the-story-so-far?r=2ri91a&utm_medium=ios&utm_campaign=post\n\n\n\nCursor\n2023-10-10\nhttps://cursor.sh/\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
"objectID": "about.html",
@@ -399,346 +406,360 @@
]
},
{
- "objectID": "lists/ai/index.html",
- "href": "lists/ai/index.html",
- "title": "Cool AI Tweets",
+ "objectID": "christianity/2023-12-13-three-little-pigs/index.html",
+ "href": "christianity/2023-12-13-three-little-pigs/index.html",
+ "title": "Three Little Pigs",
"section": "",
- "text": "TL;DR: AI is wild. Last updated per last headline. First created 2023-12-14\n\n\n\n\n\n\n\nv0.dev\n2023-12-15\nv0.dev a UI generator by vercel. React.\n\n\nCoffee by Coframe\n2023-12-15\nBuild React UIs super fast. GitHub\n\n\nAnnouncing Coffee: build and iterate on your UI 10x faster with AI ☕️👇https://t.co/0vBssgp0ue pic.twitter.com/JqwC8WpDzs\n\n— Coframe (@coframe_ai) December 13, 2023\n\n\n\n\nPrompt Engineering\nAdded 2023-12-15\nPrompt Engineering tips by OpenAI\n\n\nGenerate a Website from Screenshot\n2023-12-14\n\n\nscreenshot-to-code: upload a screenshot of any website, watch as AI progressively builds the html, iteratively improving the generated code by comparing it against the screenshot repeatedly. pic.twitter.com/TCrbJj3VS0\n\n— Siqi Chen (@blader) November 21, 2023\n\n\n\nGitHub\nhttps://screenshottocode.com/\n\n\n\ntldraw\n2023-12-14\n\n\nLooking to learn about the AI thing we made?👇Link in the next post pic.twitter.com/1FjY3eMSSF\n\n— tldraw (@tldraw) November 21, 2023\n\n\n\nhttps://makereal.tldraw.com/\nhttps://open.substack.com/pub/tldraw/p/make-real-the-story-so-far?r=2ri91a&utm_medium=ios&utm_campaign=post\n\n\n\nCursor\n2023-10-10\nhttps://cursor.sh/\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "book-reviews/2022-06-24-getting-things-done/index.html",
- "href": "book-reviews/2022-06-24-getting-things-done/index.html",
- "title": "Getting Things Done",
- "section": "",
- "text": "Principles and Core Insights\nAfter 21% my takeaways are:\n\n“My mind is weighed down by all of the ideas of things I can do.” I waste my time on todos that are just “go to doctor”, but should be “schedule appointment to go to doctor”. Anything that only takes 2 min should be done immediately. Anything longer later. But I love that clarity - I’ve often written lots of todos but not from the “next action to take” perspective.\n“Visions can be built from the ground up rather than just top down.” I’ve tried writing down big visions for my 10 year future and it never works for me. I’ve got anxiety when I try to think where I want to be in 10 years. I mean, I know I want to be happy, be best friends with my wife and kids and have meaningful relationships but whether I’m still a data scientist or a farmer by then is anyones guess. (I wanna keep my options open.) But as I look at the things on my todo list and “eventual” to do list, a different picture emerges. I know I want to be a writer (hence this blog), I know I want to travel to XYZ location, to be involved in giving back, etc. TLDR: I know things I want to experience in life. I don’t have to craft value statements and invent ideas of what to do (seems forced). Instead I can reflect on what’s currently exciting and important to me and categorize those things into high levels values/missions. This will help me when I need to prioritize so not every whim of an idea gets my attention.\nGTD book just lays out what a todo is. It breaks down the problem of doing things (big and small) into what they are (time sensitive, someday, etc). Whether I’ll follow the system is up for debate, but I love the ideas so far. It very much fits my naturally organized mind, but may not work for more spontaneous folks that don’t like as much structure and list keeping.\nMost of my “todos” have been “intended outcomes”, but not “next actions”. For example, I see I have “change car battery” on my list right now and “plan yosemite trip”. But these aren’t todos. I can’t organize and prioritize outcomes. I can only organize and prioritize actions, he says. Changing these to “call mechanic to book oil and battery change and identify an hour in your calendar”. And “open recreation.gov when walking around the house and find a campsite.”\nOrganize tasks around where you’ll do them. This is in line with Atomic habits which states to be hyper specific about where and when you’ll do something. “Call grandma when I get into my car tomorrow morning to go to work.” Or “take a moment at my desk during lunchtime to find a mechanic by opening Google Maps and plan an oil change. Find 1 open hour on the calendar.”\n\n\n\nPutting it into practice\nI’ve been using TickTick for my task manager, which has Lists (folders) where I can put my tasks and tags so I can label and categorize each task. Here’s my organization strategy:\n\nWhen jotting ideas write intended outcome and next action as my todo.\nTag with where I’ll do it and provide day or time it’s due (if applicable)\n\nMove each item to a respective list:\n\nASAP - there’s a useful urgency to this task (begin memo)\nEventually - there’s no urgency but I know I’ll want to do this\nSomeday maybe - no urgency and no commitment, but it’s an idea of something I want to do.\n\nTagging system: tag each item with either Work or Personal. Then within that, some meaningful themes:\n\nProjectName (work deliverable or personal project like “ImproveMyHealth”, “BuildApp”)1\nLocation (where I’ll do it). Around the house. At desk. In car. While cleaning the house, etc. this is habit stacking. Most work stuff is done at my desk.\nwhen I’ll do it. Stack with other activities. Examples: Before lunch.\nDuration (how long it takes: 1, 2, 4 hours, 5, 30min). This might be useful for planning.\n\nList system. Move from inbox to list when properly categorized. I like “Work” and “Personal” lists. I think since you can use unlimited tags, and tags can be nested in TickTick, there’s no substantive differ ne between list and tag. Not sure yet how to use the TickTick lists when tags are so useful. Lists of work and personal.\n\n1 Deciding Vision: As I write this, choosing a project might be a good place to insert a goal (intended outcome) as a project name. Instead of Exercise, the intended outcome is improving health. The goal is being able to run a 5k. So perhaps goal/objective is “ImproveHealth” and KR is “run 5k”. These goals can be identified by asking the 5 whys. If I have an idea if “go running” as my task, a good grounds-up vision exercise would be to ask “why do I want to run a mile tomorrow? Is it to accomplish something or to relax my mind? Why is relaxing my mind important to me? Why is my mental health important? Etc. I think subconsciously ideas come of what to do next all the time “I should work out” but rarely do I take the time to ask “why do I see the need to work out? What’s the greater vision here for this task?” Understanding that vision can help prioritize and build a desire to accomplish the task. If the idea “learn a new language” comes up, asking why can help me realize “oh, this is just for fun and won’t be for anything meatier than having fun! No pressure on this task and maybe I’ll never do it.” The nag to run a mile might help me realize “I need to take care of my health, and this can’t wait.”In summary:\n\nIdea/task comes.\nDocument in inbox.\nClarify the intended outcome and the next action.\nOrganize: tag. Set up reminder. Move out of inbox to list.\nReview and Prioritize: understand where these ideas will take you and choose your adventure.\nDo!\n\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "objectID": "christianity/2023-12-13-three-little-pigs/index.html#first-glance",
+ "href": "christianity/2023-12-13-three-little-pigs/index.html#first-glance",
+ "title": "Three Little Pigs",
+ "section": "first glance",
+ "text": "first glance\nYes this is a book review about “The Three Little Pigs”.\nThe first pig is lazy. So he builds his house of straw.\nThe second is lazy too. So he builds his house of sticks.\nThen the third is the smart one who builds his house of bricks.\nBut the first two brothers just make fun of the mason. Why waste your time?!? We’re having fun!\nSo the wolf comes and blows down the houses of the lazy brothers. They run and hide in brick brother’s house.\nMoral of the story: be lazy but have a great, well prepared brother!"
},
{
- "objectID": "book-reviews/2023-12-18-hitchhiker's-guide-to-the-galaxy/index.html",
- "href": "book-reviews/2023-12-18-hitchhiker's-guide-to-the-galaxy/index.html",
- "title": "Hitchhiker’s Guide to the Galaxy",
- "section": "",
- "text": "The book of infinite possibilities\n\n\nThis is a classic. First time reading it and I love it.\nWhat I love about the storytelling:\n\nDeclare something as unavoidably true, even if ridiculous, and build all content off of it.\n\n\nNobody cares that the infinity probability drive is absurdly improbable. But once the reader just accepts this as fact the rest of the story make sense.\n\n\nget into the mind of the reader\n\n\nsperm whale? I love when he’s detailing the whales first and last thoughts. Just so random, but hilarious.\n\n\nbe spontaneous, but connected.\n\n\nI recently watched the second season of Loki and couldn’t stand it. It was all over the place. Random here, random there. But this is great because the spontaneity flows within the confines of the story.\n\n\nlogic, with social proof, can invert the mind\n\n\nArthur dent is flabbergasted by mice being the most intelligent species. Well the logic is sound: they were just fooling us and running experiments on us the whole time so we wouldn’t think anything of it. This flip on expectation, backed by “isn’t it obvious?!” gets the reader wondering for a second before they snap back to reality: no, mice aren’t running the universe.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "objectID": "christianity/2023-12-13-three-little-pigs/index.html#not-quite",
+ "href": "christianity/2023-12-13-three-little-pigs/index.html#not-quite",
+ "title": "Three Little Pigs",
+ "section": "not quite",
+ "text": "not quite\nI’ve always been bothered by this as I read to my kids. But I’ve recently changed my opinion. Because im very much the first and second brother. We all are.\nWere inadequate. Maybe we’re not smart enough to know how to create bricks and such. Maybe we don’t have enough resources and the straw is the best we can do. And maybe we hide behind these insecurities by laughing at those who have more.\nMaybe we hide behind our insecurity and question God. The Wise One.\nSee, Christ is our Brick. He’s the safety. He’s the home we can run to when the world blows us over. He’s the Prepared One who was wishing for us to come to him the whole time. The cry of “you’ll be sorry” by this brother could perhaps be His loving warning of “please be wise. There’s a wolf out there.” Yet we don’t listen. We want to play.\nI love the third pig because he lets them in.\nAlso, the third pig never had to worry about almost getting eaten. Sure, first two enjoyed their fiddle and dance. But they almost died. The third pig is like the non-prodigal son. The one who stayed home and enjoyed the safety of the home the whole time.\nBe wise. Be forgiving. And yes, we always have a wise elder Brother who will accept us."
},
{
- "objectID": "book-reviews/2023-12-23-dune/index.html",
- "href": "book-reviews/2023-12-23-dune/index.html",
- "title": "Dune",
+ "objectID": "christianity/2023-12-24-the-christ-child/index.html",
+ "href": "christianity/2023-12-24-the-christ-child/index.html",
+ "title": "The Christ Child",
"section": "",
- "text": "shield your son too much and he’ll not grow strong enough to fulfill any destiny\ndeprogram after the machines. Let the machines do too much thinking.\nhumans must never submit to animals\n“how did the woman seal my tongue?” Paul can’t tell his father he’s going to die. Why? Sometimes our tongues are bound to speak that which can’t be changed\n“the mind commands the body and it obeys. It commands itself and it rebels”\n\nConnections: - Alan Pinkerton (killers of the flower moon) “secrets break the character of the criminal till they give it out”. Dr Yiwei cries when his scripture is given out.\n\nif I made him do it it wouldn’t be his doing.\nstorytelling: why tell us 100% of everything up front? The movie hides the secrets about Yueh.\nstorytelling: the hindsight narrator\n\nWhen God has appointed a man to die, he guides his wants to desire that place\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "The story of the Savior’s birth brings many wonderful concepts to memory and can teach us a lot about what will bring us true joy and happiness in this life.\n\nThis is my favorite video of the Christmas story. The most powerful part is the moment when Mary sings a lullaby to Jesus. I remember reading somewhere that this moment was completely impromptu by the actress, and it speaks volumes to the sacredness of the moment where Mary first laid eyes on her beloved son.\n\nSo what does this story give me perspective on at this time of year?\n\nChrist was the rightful, bloodline heir to King David’s throne. He was the King of the Jews. But he never sought authority. He never sought a throne. He instead taught “Render therefore unto Caesar the things which are Caesar’s”. He didn’t engage in turf battles. In rebellion. In insurrection. He just wanted people to find happiness in their circumstances by living with lovingkindness in their hearts for their God, their neighbor, and themselves.\nIn his role as a Savior and Judge of the world, he sought to not condemn the sinner trapped in the act of sin. Instead he came to liberate. To forgive. To support.\nIf you believe Christ was the Creator and Savior of the world, consider how He chose to come into the world. He was born in a stable, mucked with animal feces and rotting food scraps, covered with the stench of…a stable. He was rejected of a room in the inn. He was laid in a manger, perhaps a feeding trough. Maybe Mary had help giving birth, maybe she didn’t. He was born generally alone. I wonder if He chose to do his birth this way because nobody can say he had it easy. He didn’t have the special treatment of a King’s birth. No, he was born to an exhausted, weary traveler, likely on the cold hard floor of a dirty barn.\nIf we think about the Savior being the architect of this birth, think about how he planned for angels to show up to shepherds. Not King Herod. The shepherds. He would be the prophesied sacrificial Lamb. He would be the Good Shepherd. Yet these people were perhaps the minimum wage, forgotten ones outside the city gate. Yet these are those who had the privilege to see Him first. And He invited them to come and see. His whole life was an invitation.\nHe also invited the rich wise men. All are invited. He wanted all to see, and he wanted the world to know that He was for everyone regardless their circumstances. And that all of us, whether destitute or traveling with caravans, can bring the gifts of our heart to Him. We don’t need to bring gold. He just wants a “broken heart and contrite spirit.\n\nHow can we apply these things to our life, today?\n\nSeek out the forgotten. The people in our life that seem forgotten. The servants in our life (person on the other end of the customer support line, the checkout person, your supervisor) but even, your best friend who seems to be doing great…\nSeek not riches or wealth. Harod sought these things. Christ sought relationships and truth.\nDon’t seek for power. Christ deserved it. Christ had it. He turned water to wine, cursed a tree, and made a host of men fall over at the sound of his voice. Yet he didn’t touch the Romans.\nSeek to invite. To share. To uplift. Christ made shepherds feel important. He invited them to his crib side. And ” his hand is stretched out still.” (Isaiah 9:12)\n\nHe came like all of us, as a child. But he came to show us how a God would live. And he did.\n\n\n\nThe Christ as a Child, before worshipping Wise men\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "book-reviews/2023-12-19-101-essays-to-change-the-way-you-think-by-briana-weist/index.html",
- "href": "book-reviews/2023-12-19-101-essays-to-change-the-way-you-think-by-briana-weist/index.html",
- "title": "101 Essays that will Change the Way You Think, by Brianna Wiest",
+ "objectID": "img/little-lamb.png.html",
+ "href": "img/little-lamb.png.html",
+ "title": "Bryan Whiting",
"section": "",
- "text": "TL;DR: 101 ideas. This is like a modern day Meditations.\n\n\n\n\n\n\n2023-12-20 9-11\n\nhappiness of pleasure, grace (appreciation), excellence (emotional resilience)\nsome of the poorest countries are the happiest.\n\n2023-12-19 Essays 1-8\n\nwe all started as strangers\nto change your beliefs, go have experiences that manifest and confirm the beliefs you want to have\ncan only have a few truly\nour relationship with others is a pure reflection of our relationship with our self. Work on your relationship with yourself to improve your relationship with others.\ndo you look at your leg and say “that’s me?” No. That’s my leg. Do you look at your eye, heart, hair? No. Do you look at your fears, hopes or dreams and say, that’s me?\nfear and Anxiety aren’t feelings. They’re symptoms.\nWhat emotion do you feel the most often? Is it positive? If not, why does a negative feeling deserve the spotlight? What does this tell you about your relationship with yourself?\n\n2023-12-22 essays 12+\n\nknowing doing gap. “If I lived today on repeat for 10 years where will I be?” What would happen if I didn’t do X?\n\nInsights: - self actualization is what we’re all doing. We all mostly have shelter, food, job, even if minimum wage.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "book-reviews/2023-12-12-scientific-freedom/index.html",
- "href": "book-reviews/2023-12-12-scientific-freedom/index.html",
- "title": "Scientific Freedom",
+ "objectID": "index.html",
+ "href": "index.html",
+ "title": "Recent Posts",
"section": "",
- "text": "Max Plank, GPT 4\n\n\n\n“Scientific Freedom”\n\n\n\n\n\n\nTL;DR: a scientist isn’t a title, but a philosophy\n\n\n\n\n\n\nScientists are hampered by the peer review process. The notion that some tolerate the process so they can get their grants approved and get tenure, and others eventually accept the process because that’s how they got tenure, is a sad outcome of human incentives.\nInterestingly, inventions have always happened regardless of peer review. Edison didn’t need someone to review his invention to tell him whether or not it’s a valid invention. Either the light bulb illuminates or it doesn’t.\nYet that’s exactly how things go now. So what’s the result?\nPeople write to please the peer reviewer. Social conformity. Group think.\nPeople are incentivized to game the system: write a grant that gets them funding so they can get tenure and support their families.\nThe incentive to be innovative is small, though intellectual curiosity powers through for some. The incentive to stay in line and do what’s necessary to get approved is strong.\nWe’ve also learned as a consequence of 2020-2022 that a lot of scientists can indeed be wrong. Today science is becoming less about discovering truth and more about people owning their truth…people love their politics.\nWhat makes a scientist anyway? Am I a scientist because I’m a data scientist? Because I studied statistics? I’ve never published an academic article in a journal, so is that the threshold?\nI think someone who pursues truth through repeatable experimentation is a scientist. Back in Plato’s day they were just called philosophers.\n\n\nToday’s incentives\nThere’s a reason really smart people now work at Apple to invent the iPhone, Facebook to work on LLMs, and SpaceX to build rockets. The freedom they feel at these companies is empowering. Per “Scientific Freedom”, if the trends continue then people like Max Plank would never have been able to get the funding needed to invent what he did. Where’s the money? Social media. Advertising. And now, AI.\nPeople will always be inventive. But it’s by removing the social barriers that they become so.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "“All blogs are wrong, but some are useful.”\n\n\n\n\n\n\n \n \n \n Order By\n Default\n \n Title\n \n \n Date - Oldest\n \n \n Date - Newest\n \n \n \n\n\n\n\n\n\n\n\n\n\nPost Christmas blues? Remember the empty tombs\n\n\n\n\n\n\nchristianity\n\n\n\nAn empty room and an empty tomb. There is meaningful symbolism in taking down the tree. \n\n\n\n\n\nWed Dec 27, 2023\n\n\n3 min\n\n\n\n\n\n\n\n\n\n\n\n\nI’m grateful for artists\n\n\n\n\n\n\ngratitude\n\n\nchristianity\n\n\n\nThose who create, who learn, who think in new ways, who pioneer, who explore. \n\n\n\n\n\nWed Dec 27, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nAn Unknown God\n\n\n\n\n\n\nchristianity\n\n\ntheology\n\n\nfaith\n\n\ndoubts\n\n\n\nGod is indeed unknown, but that doesn’t mean He doesn’t exist. And learning His ways requires seeing all sides, being intellectually and spiritually honest, and open. \n\n\n\n\n\nWed Dec 27, 2023\n\n\n10 min\n\n\n\n\n\n\n\n\n\n\n\n\nWriting is truly enlivening\n\n\n\n\n\n\ncommunication\n\n\nwriting\n\n\n\nWrite in public. Write daily. Write about nothing. \n\n\n\n\n\nTue Dec 26, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nWrite despite AI\n\n\n\n\n\n\nai\n\n\nhumanity\n\n\n\nI’m fully aware an AI is listening and all that I write just trains it. Perhaps this is a perversion of my work (if I made money off it). Why write if everything I write just trains an AI agent to do their job better? \n\n\n\n\n\nTue Dec 26, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nThings I’d like in a blog\n\n\n\n\n\n\nproductivity\n\n\nquarto\n\n\n\nWhat quarto is great at and what I’d like out of it. \n\n\n\n\n\nTue Dec 26, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nI’m grateful for blankets\n\n\n\n\n\n\ngratitude\n\n\n\nMy grandma made blankets cool \n\n\n\n\n\nTue Dec 26, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nHow to blog from your phone using quarto (or hugo/jekyll/etc) using Working Copy and Obsidian\n\n\n\n\n\n\nproductivity\n\n\ntech\n\n\ntutorial\n\n\n\nWorking copy is $20, obsidian is free (and optional). I’ll show you how to set this up (iPhone only I think?) \n\n\n\n\n\nTue Dec 26, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nI’m grateful for doctors\n\n\n\n\n\n\ngratitude\n\n\n\nDoctors of the body. Doctors of the mind. Doctors of the soul…Any engaged in the pursuit of healing. \n\n\n\n\n\nMon Dec 25, 2023\n\n\n3 min\n\n\n\n\n\n\n\n\n\n\n\n\nAI Robots are mastering tasks faster than humans\n\n\n\n\n\n\nai\n\n\ntech\n\n\nfuturism\n\n\n\nFrom walking to labyrinth, the future looks bright for these machines. The key is the objective functions, time, and training data. \n\n\n\n\n\nMon Dec 25, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nTwas the Night before Christmas\n\n\n\n\n\n\nchristmas\n\n\nask gpt\n\n\n\nThe origin of the story, and a link to the 1912 publication. \n\n\n\n\n\nSun Dec 24, 2023\n\n\n3 min\n\n\n\n\n\n\n\n\n\n\n\n\nThe Christ Child\n\n\n\n\n\n\nchristianity\n\n\n\nThe birth of Jesus Christ teach us what we can do to have greater peace in this life \n\n\n\n\n\nSun Dec 24, 2023\n\n\n4 min\n\n\n\n\n\n\n\n\n\n\n\n\nLight\n\n\n\n\n\n\ngratitude\n\n\n\nI set up 72 lights today. \n\n\n\n\n\nSun Dec 24, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nHis is the light that we should share\n\n\n\n\n\n\nchristianity\n\n\n\nWho is the source of light? \n\n\n\n\n\nSun Dec 24, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nThe Best of Edgar Allen Poe\n\n\n\n\n\n\nbook review\n\n\n\nYou think me mad? \n\n\n\n\n\nSat Dec 23, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nDune\n\n\n\n\n\n\nbook review\n\n\n\nThe plans and risks of generations \n\n\n\n\n\nSat Dec 23, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nThe Killers of the Flower Moon\n\n\n\n\n\n\nbook review\n\n\n\n\n\n\n\n\n\nFri Dec 22, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nThe Allegory of the Dodo\n\n\n\n\n\n\nproductivity\n\n\nbusiness\n\n\nkids books\n\n\n\nWhen in having solved a problem we inadvertently create problems for ourselves \n\n\n\n\n\nThu Dec 21, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nGitHub copilot levels up\n\n\n\n\n\n\nai\n\n\nproductivity\n\n\ntech\n\n\n\nWhat will it do next? \n\n\n\n\n\nThu Dec 21, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nThe Little Red Hen\n\n\n\n\n\n\nbook review\n\n\nstartups\n\n\n\nPeople love to enjoy the fruits, but who will bake the bread? \n\n\n\n\n\nWed Dec 20, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nHow to encourage your kids\n\n\n\n\n\n\nparenting\n\n\n\nUnexpected advice on how to be a great parent when fostering talent \n\n\n\n\n\nTue Dec 19, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\n101 Essays that will Change the Way You Think, by Brianna Wiest\n\n\n\n\n\n\nbook review\n\n\nphilosophy\n\n\nmindfulness\n\n\n\nWe all started as strangers. It’s a choice who remains one. \n\n\n\n\n\nTue Dec 19, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nHitchhiker’s Guide to the Galaxy\n\n\n\n\n\n\nbook review\n\n\nstorytelling\n\n\n\nHow did it take me so long to finally read this? \n\n\n\n\n\nMon Dec 18, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nChristmas celebrates a Living Christ\n\n\n\n\n\n\nchristianity\n\n\nai\n\n\n\nChrist is the way, the light and life of the world. Always has been. Always will be. \n\n\n\n\n\nSun Dec 17, 2023\n\n\n3 min\n\n\n\n\n\n\n\n\n\n\n\n\nWhat is humanity\n\n\n\n\n\n\nai\n\n\n\nHumanity seeks to resolve problems. What happens when they’re too easily solved? \n\n\n\n\n\nSat Dec 16, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nPrompt techniques\n\n\n\n\n\n\nai\n\n\n\nOpen AI strategies \n\n\n\n\n\nFri Dec 15, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nLLMs can find a needle in the haystack\n\n\n\n\n\n\nai\n\n\n\nGPT outperforms Claude. \n\n\n\n\n\nFri Dec 15, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nThree Little Pigs\n\n\n\n\n\n\nbook review\n\n\nchristianity\n\n\n\nA story with a terrible but incredibly relatable morale \n\n\n\n\n\nWed Dec 13, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nLearn by doing\n\n\n\n\n\n\ndata science\n\n\ntech\n\n\n\nYou don’t need a PhD. Or a masters. You just need to hustle \n\n\n\n\n\nWed Dec 13, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nScientific Freedom\n\n\n\n\n\n\nbook review\n\n\n\nInnovation comes by varied incentives: intellectual curiosity and economic incentive \n\n\n\n\n\nTue Dec 12, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nFuture of Coding\n\n\n\n\n\n\ndata science\n\n\n\nAs our jobs are automatable, we need to discover what’s not \n\n\n\n\n\nTue Dec 12, 2023\n\n\n4 min\n\n\n\n\n\n\n\n\n\n\n\n\nHow do you want to be remembered?\n\n\n\n\n\n\nleadership\n\n\nservice\n\n\n\nIt’s a question as old as time, but misses the mark and is distracting from the more important question. \n\n\n\n\n\nFri Dec 23, 2022\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nBuilding a complex R Shiny Dashboard Using ChatGPT\n\n\n\n\n\n\ndata science\n\n\nr\n\n\nshiny\n\n\nfuturism\n\n\nchatgpt\n\n\n\nChatGPT can create amazing boilerplate code and can point you in the right direction, but it cannot debug data problems easily (yet). It can read in data from the internet, make time-series forecasts, and visualize the data (with a little assistance). \n\n\n\n\n\nThu Dec 15, 2022\n\n\n8 min\n\n\n\n\n\n\n\n\n\n\n\n\nModels as a Service and the Future of Data Science\n\n\n\n\n\n\ntech\n\n\nstartups\n\n\nbusiness\n\n\ndata science\n\n\n\nData Scientists shouldn’t be scared their jobs will disappear. They should be excited that they’ll be joining companies that automate what they’re currently doing and offer it as a service. \n\n\n\n\n\nTue Dec 6, 2022\n\n\n4 min\n\n\n\n\n\n\n\n\n\n\n\n\nWriting a Tech Resume\n\n\n\n\n\n\ncommunication\n\n\n\nHow to write a resume to get into a FAANG company. \n\n\n\n\n\nFri Nov 18, 2022\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nResume Substance over Style\n\n\n\n\n\n\ncommunication\n\n\n\nThe content of your resume is more important than the design. \n\n\n\n\n\nFri Nov 18, 2022\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nWhat is a model?\n\n\n\n\n\n\ndata science\n\n\n\nPutting the mysterious in context. \n\n\n\n\n\nSat Nov 12, 2022\n\n\n5 min\n\n\n\n\n\n\n\n\n\n\n\n\nDe-sciencing Data Science and Talking Like a Normal Person\n\n\n\n\n\n\ndata science\n\n\ncommunication\n\n\nprinciples\n\n\n\nHow can data scientists bring their technical knowledge to a non-technical audience? Here are my lessons learned from seven years in the data trenches. \n\n\n\n\n\nWed Jul 20, 2022\n\n\n6 min\n\n\n\n\n\n\n\n\n\n\n\n\nWhen data collection goes too far\n\n\n\n\n\n\nproductivity\n\n\nfuturism\n\n\ntech\n\n\nsurveillance\n\n\n\nIt’s important for us to track certain things about ourselves. But this is clearly not exactly healthy IMO. \n\n\n\n\n\nSun Jul 3, 2022\n\n\n4 min\n\n\n\n\n\n\n\n\n\n\n\n\nPomodoro Principles\n\n\n\n\n\n\nproductivity\n\n\nticktick\n\n\n\nThe art of focus through 25min work blocks. \n\n\n\n\n\nTue Jun 28, 2022\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nIncrease Focus through Writing Down Everything\n\n\n\n\n\n\nproductivity\n\n\nmusings\n\n\nticktick\n\n\n\nTo be more present, get it all down and into a system. \n\n\n\n\n\nTue Jun 28, 2022\n\n\n3 min\n\n\n\n\n\n\n\n\n\n\n\n\nHow I use TickTick\n\n\n\n\n\n\nproductivity\n\n\n\nAn overview of how I apply the principles of “Getting Things Done”, or GTD. \n\n\n\n\n\nSat Jun 25, 2022\n\n\n7 min\n\n\n\n\n\n\n\n\n\n\n\n\nGetting Things Done\n\n\n\n\n\n\nbook review\n\n\nproductivity\n\n\n\nA powerful system to be more present. \n\n\n\n\n\nFri Jun 24, 2022\n\n\n6 min\n\n\n\n\n\n\n\n\n\n\n\n\nWhat is the best todo list manager?\n\n\n\n\n\n\nproductivity\n\n\napp review\n\n\n\nI’ve reviewed 20+ apps. I think I’m done reviewing them. \n\n\n\n\n\nThu Jun 23, 2022\n\n\n11 min\n\n\n\n\n\n\n\n\n\n\n\n\nNew ideas are distractingly exciting\n\n\n\n\n\n\nproductivity\n\n\npriorities\n\n\nideation\n\n\n\nBe careful of recency bias with your ideas. \n\n\n\n\n\nThu Jun 23, 2022\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nDownstream Impact\n\n\n\n\n\n\ndata science\n\n\nexperimentation\n\n\n\nWhen you do an online experiment, you’re changing the future forever. \n\n\n\n\n\nWed Jun 22, 2022\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nThe Power of Party Pay by Visible\n\n\n\n\n\n\nstartups\n\n\nstrategy\n\n\n\nVisible wireless, owned by Verizon, offers users $25 off per month if they join a meaningless party. What’s the effect? \n\n\n\n\n\nTue Jun 21, 2022\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nProducts for dads\n\n\n\n\n\n\nproductivity\n\n\nfatherhood\n\n\n\nWhat products am I still using 5 years later? \n\n\n\n\n\nTue Jun 21, 2022\n\n\n6 min\n\n\n\n\n\n\n\n\n\n\n\n\nBlogging with Quarto, and why I don’t use Medium anymore\n\n\n\n\n\n\ndata science\n\n\n\nI’m trying out quarto and I like it. \n\n\n\n\n\nMon Jun 20, 2022\n\n\n4 min\n\n\n\n\n\n\n\n\n\n\n\n\nThe Author’s Angle Matters\n\n\n\n\n\n\ncommunication\n\n\nwriting\n\n\n\n\n\n\n\n\n\nSat Jul 17, 2021\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nProductivity notes\n\n\n\n\n\n\ndata science\n\n\nproductivity\n\n\n\nRunning list of productivity ideas. \n\n\n\n\n\nSat Dec 19, 2020\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nHow to mock the value of an ML solution\n\n\n\n\n\n\ndata science\n\n\ninnovation\n\n\n\nSell your solution before training your model. \n\n\n\n\n\nWed Dec 16, 2020\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nGet rid of the todo list. Calendar everything.\n\n\n\n\n\n\nproductivity\n\n\n\nIf you’re not willing to block time for it, does it event matter? \n\n\n\n\n\nTue Dec 15, 2020\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nData Beats Opinions\n\n\n\n\n\n\ndata science\n\n\ndecisions\n\n\n\nOpinions from the customer shape the product. Their behavior (captured as data) speaks louder than their words. \n\n\n\n\n\nTue Dec 15, 2020\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nSemantic Versioning for Data Science Models\n\n\n\n\n\n\ndata science\n\n\nmachine learning\n\n\n\nMeaningful versioning for data science models and machine learning pipelines. \n\n\n\n\n\nMon Jul 2, 2018\n\n\n5 min\n\n\n\n\n\n\n\n\n\n\n\n\nBuying a used car the data science way: Part 2\n\n\n\n\n\n\nwebscraping\n\n\nr\n\n\nregression\n\n\nanalysis\n\n\npricing\n\n\nbuying things\n\n\n\nHow I analyze used car data to find under-valued cars, and why none of my analysis matters.\n\n\n\n\n\nSun Feb 19, 2017\n\n\n14 min\n\n\n\n\n\n\n\n\n\n\n\n\nBuying a used car the data science way: Part 1\n\n\n\n\n\n\nwebscraping\n\n\ncars\n\n\npricing\n\n\nr\n\n\n\nHow I scrape used car data.\n\n\n\n\n\nSat Feb 18, 2017\n\n\n4 min\n\n\n\n\n\n\n\n\n\n\n\n\nThe DataViz battle: Plotly vs ggplot2\n\n\n\n\n\n\ndataviz\n\n\nr\n\n\nggplot\n\n\nplotly\n\n\n\nWhat can you do with plotly vs. ggplot2 and how do they compare on a simple chart?\n\n\n\n\n\nFri Feb 10, 2017\n\n\n7 min\n\n\n\n\n\n\n\n\n\n\n\n\nStarting with Jekyll\n\n\n\n\n\n\nproductivity\n\n\nblogging\n\n\njekyll\n\n\n\nHello world.\n\n\n\n\n\nThu Nov 17, 2016\n\n\n5 min\n\n\n\n\n\n\nNo matching items\n\n\n \n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only.",
+ "crumbs": [
+ "Recent Posts"
+ ]
},
{
- "objectID": "book-reviews/2023-12-23-the-best-of-edgar-allen-poe/index.html",
- "href": "book-reviews/2023-12-23-the-best-of-edgar-allen-poe/index.html",
- "title": "The Best of Edgar Allen Poe",
+ "objectID": "gratitude/grateful-for-artists.html",
+ "href": "gratitude/grateful-for-artists.html",
+ "title": "I’m grateful for artists",
"section": "",
- "text": "Raw Notes\nThe Tell-Tale Heart:\n\nyou think me mad. Nah, I’m cool. But the man grows increasingly mad.\nguilt follows the commission, before then the guilt is only in the hesrt. After it consumes.\n\nThe Cask of Amontillado:\n\nwatch your pride. Don’t let it trick you into walking into catacombs\n\nThe Masque of the Red Death:\n\nfruitless to avoid it\nallegory of phases of life\nPrince Prospero tries to fight death, but dies in the process\n\nThe Raven\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Someone sent me this video:\n\nThis artist has 536 followers on YouTube. Not exactly an influencer.\nWhat makes someone special? Because other people think they’re special? Because they have a lot of followers?\nThis video has some cool insights, things I never considered before. The image is really meaningful.\n\n\n\nLittle Lamb, by Jenedy Paige\n\n\nIm grateful for people who do research. Who create. Who take on projects they don’t want to but feel inspired to take on. For people who connect the dots in new ways.\nCreativity is a process. Everyone has the capacity to be creative. Everyone has uniquely experienced a unique combination of things. Therefore, when they synthesize those things, they become like a little Da Vinci.\nNo, this isn’t the Mona Lisa. But it doesn’t have to be to be meaningful and powerful. Even if it only reaches 536 YouTube subscribers.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "book-reviews/2023-12-22-the-killers-of-the-flower-moon/index.html",
- "href": "book-reviews/2023-12-22-the-killers-of-the-flower-moon/index.html",
- "title": "The Killers of the Flower Moon",
- "section": "",
- "text": "TL;DR:\n\n\n\n\n\n\nOn Spotify\n\nHale: self made man. Cowboy. Learned the trade. Got one cow. Then many. Then had 25,000 acres.\nAlly. Used his position as a white guy to advocate on behalf of the Osage.\n\nCherokee nation parceled and “first come first serve” over 42,000 acres\nWild West: private eye 1921 > than local sheriff\n1850 Alan Pinkerton - private eye.\nguardians - Osage couldn’t spend their own money without approval\ncriminal’s secret eats away at him\njudicial execution is a challenge of taming the evils within even those on the “right side of the law” calling for justice. Do we really need to execute someone? What if they’re actually innocent?\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "objectID": "gratitude/2023-12-25-grateful-for-doctors/index.html",
+ "href": "gratitude/2023-12-25-grateful-for-doctors/index.html",
+ "title": "I’m grateful for doctors",
+ "section": "",
+ "text": "The Healer’s Art\nI didn’t break my first bone until I was 17, but I visited my first doctor many years before.\nI grew up with an odd fever syndrome that came and went every six weeks. 105 degree fevers each time. I remember two distinct moments hallucinating that the room was splitting and shifting and another when bees were covering the corner of my bunk. The delirium was real.\nEventually it went away. But there was also a doctor there each time I got sick. One for me, to take my blood. One for my mom, to try to comfort and counsel her.\nI’m grateful for doctors of the mind who help alleviate sicknesses of stress, anxiety, depression, etc. I know more people with some type of recurring emotional distress than I know without. I see the benefits reaped by those who seek emotional healing through therapy. I’m proud of the friends who struggle, but who don’t struggle alone. The only way out, is with.\nI’m grateful for healers of the soul. Christ invited everyone to change their ways by letting go of the things that didn’t bring them true happiness, and focus on the things that do. In one invitation, he urged us to come unto him “that I may heal you”. I’ve witnessed spiritual healing through prayer, which is the process of going from feeling forgotten to then knowing I am a child of God and he has a plan for me.\n\n\nHe Came to Heal\nThis to me is the meaning of Christmas: healing. Freedom from bruised bodies, broken minds, or hurting hearts. He came to be broken, so that we may be healed.\n\nBut he was wounded for our transgressions, he was bruised for our iniquities: the chastisement of our peace was upon him; and with his stripes we are healed. Isaiah 53:5\n\n\nHe will take upon him their infirmities, that his bowels may be filled with mercy, according to the flesh, that he may know according to the flesh how to succor his people according to their infirmities. Alma 7:12\n\nNot all healing happens in an instant, and sometimes some things don’t fully heal or heal at all. But I’m grateful healing is possible and there are doctors in the constant pursuit of it. And I’m grateful for friends who have, at times, helped heal me in all these ways.\nMost of us won’t be known as doctors. But all of us try the best we can to help heal our loved ones, such as healing a child with a band aid, a spouse with a hug, a friend with a lunch. Thank you for doing the work of Christmas, all year round.\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "book-reviews/2023-12-20-the-little-red-hen/index.html",
- "href": "book-reviews/2023-12-20-the-little-red-hen/index.html",
- "title": "The Little Red Hen",
+ "objectID": "posts/2023-12-15-llms-can-find-a-needle-in-the-haystack/index.html",
+ "href": "posts/2023-12-15-llms-can-find-a-needle-in-the-haystack/index.html",
+ "title": "LLMs can find a needle in the haystack",
"section": "",
- "text": "TL;DR: Be the hen. Do work!\n\n\n\n\n\n\n“The Little Red Hen” is a classic folk tale that tells the story of a hardworking red hen who finds wheat seeds and decides to plant them. Throughout the story, she asks other farm animals for help with various tasks—planting the wheat, harvesting it, milling the flour, and baking the bread. Each time, the animals refuse to help. However, once the bread is baked and the hen asks who will help her eat it, all the animals eagerly volunteer. She declines their offers, choosing to eat the bread with her chicks, who assisted her in the work.\nSo, too, do startups everywhere struggle to get funding. 100 rejections, etc.\nAnd great products don’t get attention until there’s social proof. Many e-commerce companies understand this - sales take off as more people buy, which gets more reviews, which gets more trust - even if the product didn’t change much between first sale to the last.\nThe only difference between startups and the red hen is she perhaps had a recipe. She didn’t have to wonder, “is baking bread possible? Will the recipe be delicious?”\nIt takes a founder’s belief to see what isn’t yet so and build it anyway. Be the founder. Be the hen. Dream big. Go bake some bread even if nobody’s watching.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Is RAG necessary when you have incredible memory?\n\n\n\n\n\n\n\nContext\nCheck out this thread:\n\n\n(1/8) The Needle in the Haystack done by @GregKamradt was an amazing analysis of retrieval performance! Greg has graciously allowed us to build on his work with a repository that is now OSS.@natfriedman We have a much more rigorous test we’ve put out based on this idea.… pic.twitter.com/i5O8zrcwQT\n\n— Aparna Dhinakaran (@aparnadhinak) December 15, 2023\n\n\nThis is a powerful analysis. Sure, Anthropic will find a way to improve or challenge the results. But the point is clear: these technologies can remember hyper specific 7-digit random numbers out of a batch of 126,000 tokens, where a token is roughly 4 characters. GPT is clear winner here, too.\nAlso, open source is getting incredibly good. This implies the future is open source.\n\n\nComparing @OpenAI #GPT4 Turbo to @MistralAI GPT-4 is pretty good in that region in general. Interesting to see how @MistralAI scales to larger context windows pic.twitter.com/WQo6MmGIHh\n\n— Aparna Dhinakaran (@aparnadhinak) December 15, 2023\n\n\n\n\nImpact\nRAG can be used to make retrieval more efficient. But if retrieval is already super efficient maybe RAG is only a short term thing. Context lengths of 10m tokens…probably by next year right?\nStart of the year we were at 4K tokens. Now there are 126,000 tokens. 30x improvement. So to do another 30x improvement is 3.76M. So yea, by next year you should be able to just load the entire RAG database into memory. But…gonna be super expensive.\nPoint is: would GPT be this effective if it was using RAG over a database? Or is it more effective loading it all into context?\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "books.html",
- "href": "books.html",
- "title": "Book Reviews/Notes",
+ "objectID": "posts/2022-06-25-how-i-use-ticktick/index.html",
+ "href": "posts/2022-06-25-how-i-use-ticktick/index.html",
+ "title": "How I use TickTick",
"section": "",
- "text": "Goal: To spend less time consuming more, more time consuming intentionally.\nMostly incoherent bullet points of thought provoking moments.\n\n\n\n\n\n\n\n\n\n\n\n\n101 Essays that will Change the Way You Think, by Brianna Wiest\n\n\nWe all started as strangers. It’s a choice who remains one.\n\n\n\n\n\n\nDec 19, 2023\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nDune\n\n\nThe plans and risks of generations\n\n\n\n\n\n\nDec 23, 2023\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nGetting Things Done\n\n\nA powerful system to be more present.\n\n\n\n\n\n\nJun 24, 2022\n\n\n\n\n\n\n\n\n\n\n\n\nHitchhiker’s Guide to the Galaxy\n\n\nHow did it take me so long to finally read this?\n\n\n\n\n\n\nDec 18, 2023\n\n\n\n\n\n\n\n\n\n\n\n\nScientific Freedom\n\n\nInnovation comes by varied incentives: intellectual curiosity and economic incentive\n\n\n\n\n\n\nDec 12, 2023\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nThe Best of Edgar Allen Poe\n\n\nYou think me mad?\n\n\n\n\n\n\nDec 23, 2023\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nThe Killers of the Flower Moon\n\n\n\n\n\n\n\n\n\nDec 22, 2023\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nThe Little Red Hen\n\n\nPeople love to enjoy the fruits, but who will bake the bread?\n\n\n\n\n\n\nDec 20, 2023\n\n\n\n\n\n\n\n\nNo matching items\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only.",
- "crumbs": [
- "Book Reviews/Notes"
- ]
+ "text": "I’ve recently read “Atomic Habits”, “Deep Work”, “The Power of Habit”, “Getting Things Done”, “Essentialism” and “Effortless”. These productivity paradigms influence this post.\nInspired by this Reddit post, I’ve tried thinking how I want apply the principles of GTD and these other paradigms. I think GTD is great for me because it helps me get ideas out of my head and Essentialism is a better pursuit to make sure I’m not caught up in minutia. I don’t see them as incongruent, but one as an ideation organization approach and the other as a relentless focus.\nFive steps of GTD:\n\nCapture\nClarify\nOrganize\nReview\nEngage\n\n\nCapture\nI have an iOS shortcut that lets me quickly add a TickTick task to my inbox. I put this on my Home Screen for instant access and to avoid distraction. I want to capture the idea, not open TickTick and get distracted by other ideas.\nOr I use the TickTick today widget and upload it to today directly. Depends on the task and how detailed I want to be.\n\n\nClarify\nHere’s the system I use to clarify.\nDefinitions:\n\nTask: an action item.\nProject: an intended outcome with multiple tasks.\nEpic: an intended outcome with multiple projects.\nKey Result: high level, long term accomplishment. (Increase revenue 10%)\nObjective: a squishy, feel-good purpose statement. “Deepen relationships with loved ones.”\n\nTask grooming process: Groom the inbox every night and identify the intended outcome and next action. Tag the task.\n\nIf the task was “send Birthday card”, that’s really an intended outcome and is a small project. The tasks are 1) buy card from CVS, 2) put in envelope with stamp, 3) drop in mailbox. These three subtasks need to be understood. But a separate inbox entry may be “pay phone bill by sitting down at my computer at 10pm”. This isn’t a project. It’s just a one off thing. But it can still be clarified with details.\nLabel if it’s a project or not. Use iOS shortcuts to make this really easy. For example, gg = “🎯 #Project”\nThe challenge is if I should have things as projects or as individual tasks. I would say that as much as things can be independent, keep them tasks. Like “apply for credit card1” and “apply for credit card 2” are both personal finance tasks. But they are independent.\nA task isn’t necessarily short and a project isn’t necessarily long. It’s just whether this Indy ended outcome has multiple steps or not that might be done over a few days, whereas a task is probably done in one sitting.\nIt’s assumed that unless something is a task, it’s a project.\nEvery epic has its own name. I nest epics under “Work > #Epic1, #Epic2”, etc. This helps me organize my work deliverables. I have like 5 epics right now at work that will span several quarters.\n\nThen I may choose to add tags with metadata, which I haven’t figured out if this is useful yet. But “Atomic Habits” says the best way to get stuff done is to stack it. So here is my tagging system to help me stack:\n\nCommitment: ASAP, Eventually, Someday Maybe. Will I do this task eventually? Pay a bill - yes. Learn a language - someday maybe.\n\nDuration: 5min, 30min, 1hr, 2hr, 4hr. 9hr (all day)\nLocation (where to do the task): Home, Desk, Car, Backyard. Most work things are done at my desk. But perhaps some can be done while taking a walk. Most personal things are done around the house or on the couch at night.\nTime of Day: “1: Pre work”. “2: start of workday. 3: lunchtime, 4: afternoon, 5: after work, 6: nighttime”\nFocus: “Deepwork” or “shallow”. Deep work should be uninterrupted and be 3 hours long.\nSeverity: S1: if I don’t do this nothing will happen. S3: not doing this will have bad consequences or lots of regret. S2 - in between :).\nDeadline: I’m still thinking about this one. To me there’s a difference between mowing the law which has a linear cost of looking worse each day for not being cut and failing to pay my taxes, which has festeringly worse consequences the longer I put it off (pay a fine, end up in jail, etc.). I guess this is captured implicitly in “ASAP” commitment. But paying my taxes is only a cost (preventing negative outcomes) versus the ASAP of getting my project finished (positive consequences).\n\nNone (default): no consequence for never doing this.\nBinary: You either make the deadline or never and this opportunity ceases forever.\nRolling: If you miss the deadline there will be another.\nFestering: the longer you put this off the worse it becomes, but no deadline.\n\n\nAs I write this, this sounds like overkill…classic me. I sincerely doubt it’s useful to of all of this. But having the system available will let me use what I need when I need it. If I’m still using any of this in a month that’ll be a miracle. But I’m going to start with something in mind and go from there.\n\n\nOrganize\n\nOnce clarified, move out of inbox into a list “work” or “personal”.1\nPrioritize.\nDelegate. (Myself, someone else.)\nAssign due dates and times (if not already done)\n\n1 Tags and lists are identical except Lists have sections and views (such as Kanban). In a list you can sort by tag and vice versa. But each task can only appear in one list whereas tags are infinite. I noticed that choosing which list something belongs to can be cumbersome if it’s in a nested list, so I want only high level lists (no folders) and have this be simple. I also have “personal” and “work” tags which have projects within them, so I don’t really see why I’d need to duplicate this. On second thought, each task can only belong to one work project so it’d make sense for this to be my grouping. I suppose I just don’t want to manage all that. Or, I chose to use tags to manage my projects because I was cheap and I can only have 9 lists on the free version but unlimited tags? Either way this is easy to reorganize later if I choose to switch. Tags it is for now.\n\nReview\n\nEach night I will review the days tasks and re-prioritize as needed.\nPin the one task/project I want to get done that day.\n\nEssentialism says to take the 6 tasks of the day and choose just one. Then identify what else you need to say no to to ensure you have the energy to accomplish that task.\n\nWeekly planning session to review all of this.\n\n\n\nEngage\n\nDon’t work on any tasks in a day/week that weren’t planned for that day (unless something comes up).\n\n\n\nFinal Thoughts\nA task system should be saving me time, not consuming it. It should be relaxing me and my anxieties, not amplifying them. I’m going to give this 30 days and see if my stress decreases and productivity increase. If so, I’ll keep using this system.\nGTD states that by not writing down all the ideas, you waste energy because things come back to nag you later. But surely energy is wasted by organizing a valueless idea.\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "christianity/2023-12-24-he-is-the-light-that-we-should-share/index.html",
- "href": "christianity/2023-12-24-he-is-the-light-that-we-should-share/index.html",
- "title": "His is the light that we should share",
+ "objectID": "posts/2022-07-20-de-sciencing-data-science-and-talking-like-a-normal-person/index.html",
+ "href": "posts/2022-07-20-de-sciencing-data-science-and-talking-like-a-normal-person/index.html",
+ "title": "De-sciencing Data Science and Talking Like a Normal Person",
"section": "",
- "text": "When Christ was teaching his disciples, he invited them to share their light to the world. But then he elaborated in an interesting way:\n\n3rd Nephi 18:24 Therefore, hold up your light that it may shine unto the world. Behold I am the light which ye shall hold up—that which ye have seen me do. Behold ye see that I have prayed unto the Father, and ye all have witnessed.\n\nThe Lord is their light. He is our light. He is the light we hold up to the world and let shine. He is our light that we let shine.\nThe pronouns here are impactful by their implication:\n\nHe assumes we have light\nHe assumes we have Him as our light\nHe invites us to share the light we possess\nThe light we possess is His light\nBut it is ours. And it is his. Our light is His light. His light is our light.\n\n“Let your light so shine before men that they may see your good works…”. Read in this new way: obtain the Lord as your light. Then shine His light before men.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Building a Data Science culture\n\n\n\n\n\n\nTL;DR: Always keep it simple and always bring it back to the business need.\n\n\n\nA coworker once asked me for advice on how to bring technical rigor into a non-technical culture.\n\n\nWhen the Simple Solution Wins\nA few years back I wanted to see if I could predict which NBA team would win in their next match. FiveThirtyEight is a blog that does this every day. Could I build a model as good as theirs?\nI built a few models and put up a site. Below I have a table of accuracies for each of my models compared to FoveThirtyEight’s.\n\nWhat I learned is that Nate Silver’s FiveThirtyEight three models are only 66% accurate. In 20 hours I got a model as accurate (V02 above) as his using just team level data (win %, recent win streak, etc.) and no player data (injuries, etc.). Also my model was built using data from 2005-2017 to predict 2019 performance. His model was updated daily with recent stats.\nBut even more glaring is that if I made a prediction on a single feature (Home % > Away %) I could achieve 63.7% accuracy. This feature is a 1 if the home teams win % is greater than the away teams win %, 0 otherwise. Meaning, the home team wins 63.7% of the time their season win % is higher than the away teams season win %.\nWas it worth building an XGBoost model with 30 features to get 65.8% accuracy when just a simple calculation is 63.7% accurate? Depends on business context and how valuable that 2% gain is and what the development/engineering cost of deploying a model is. But if I were you I’d start with the simple model to get it into production to start generating value and then come back later to optimize.\nCaveat: it took a lot of work to discover that feature. 90% of the problem was building the data set. So at the end, building a model was pretty simple once I had the data. (This is where cloud AutoML will take over going forward, sorry!.) And I needed the model to see the feature importance (this was the top feature.) This is a valid data science approach: use modeling to discover the patterns. But just because you have a model doesn’t mean it’s worth deploying.\nKeep it business focused: when building a recommendation system, start with “what’s most popular”. Why would you need K-means or whatever to begin? And can’t you build your own K-means by creating heuristic filters? For example: if you want to recommend me a shirt based on shirts I’ve searched, you can find all shirts of same color, price, and size and recommend those. Building a model to take into account dimensions, patterns, what others buy, etc is fine but should be a V2.\n\n\nMindsets\nThere’s a temptation to act like an academic professor and use fancy jargon when working with your coworkers. Even worse, there’s a temptation to think you’re smarter than your coworkers because you know this mathy thing that they don’t. But remember this quote:\n\nWhen they are learned they think they are wise…1\n1 2 Nephi 9:28\nThis pride power struggle is your downfall for two reasons:\n\nIt blinds you to thinking your solution is good just because someone else can’t understand it.\nIt blinds you from focusing on what matters most: getting the simplest and best solution.\n\nYour coworkers aren’t your students eating you. They’re your smart coworkers solving the same problems you are.\nA better mindset is that of a professional athlete, like a football player. They might be dumb as bricks at times but they’re all incredibly smart players. Can you explain your technical solution like a pro football player would?\n\nBad: “I built an XGBoost classification model that reduces the RMSE by minimizing the Gini coefficient on each node…”\n\nBetter: “I noticed that most users close the app after they reach this page. I tried looking at some reasons why and found a few common behaviors (DISCUSS, get feedback). I think we can predict when this will happen and here’s a framework to do that.”\n\nThere’s a temptation to think “I did all this technical work, I should get credit for it!” My manager once coached me, “Your stakeholders should be amazed by how simple you made it. Not by how complicated the problem is.”\nThe best example I know of this is Chris Voss, an FBI negotiator. Listen to his podcasts or YouTubes. He’s a humble guy and doesn’t get into the technical stuff. He knows the principles and keeps the explanations simple. He helps you feel smart when you listen to him. He’s one of the best negotiators in the world and if you were having dinner with him you’d think he was a normal guy. This is how data scientists should be.\n\n\nBugs\nThere are two types of bugs in data science: Software bugs and data bugs. You’ll spend 80% of your ramp up time on a project setting up your environment, learning the data, trying to understand the primary keys and logging and unique values, etc.\n\n\n\n\n\n\nNobody cares about how you do any of this (the same way you don’t care about how cereal ends up in your grocery store), but they do care that you do it because it shows that you’re thinking about the right things.\nThey’ll also worry that you’re spending too much time on the wrong thing, so make sure you believe your actions are solving a business purpose.\n\n\n\n\n\nTakeaways\n\nFocus on principles, not technical details.\nGet the simplest, useful solution into production first to start creating value. Then figure out if it’s worth optimizing. If your product gets deprecated in 3 months then the fancy model will be wasted time anyway.\nYour stakeholders should be amazed by how simple you made it. Not by how complicated the problem is.\nIf you can’t explain a trend with a simple bar/line chart, a fancy regression model won’t do.\nStart with the non-technical (basic math) solution. Then see if you can beat the basic math with a fancy model. Evaluate the cost/benefit of the technical solution relative to the dumb version.\nwhen you have a fancy solution, make sure you compare it to the basic solution. Make sure you can beat the solution your teammates come up with. And make sure you carefully lead them from “their” solution to yours. “At first I tried X. X was fine. But X didn’t account for Y. So I built a model M and saw that M out performs X by 10%.”\nJust because you know/use math, doesn’t mean it’s right for the business. Data Scientists can easily be myopic because the work we do is challenging and deep in the weeds.\nYour boss doesn’t care about the nature of the bugs, but you should try to bring first principles to the table so she can cover for you. “I keep hitting this bug in the data. Can we ask/work with Infra team to solve this because it eats my time.”\n\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "christianity/2023-12-17-christmas-celebrates-a-living-christ/index.html",
- "href": "christianity/2023-12-17-christmas-celebrates-a-living-christ/index.html",
- "title": "Christmas celebrates a Living Christ",
+ "objectID": "posts/2023-12-24-twas-the-night-before-christmas/index.html",
+ "href": "posts/2023-12-24-twas-the-night-before-christmas/index.html",
+ "title": "Twas the Night before Christmas",
"section": "",
- "text": "Away in a manger.\n\n\n\nOld news\nElder Neil Anderson’s invitation here is worth listening to. It’s an invitation to ponder the reality of the birth of our Savior. It’s rather simple, but it evokes a profound question we should all consider.\n\n\nHow wonderful it is to celebrate the birth of our Savior and Redeemer. Jesus Christ is the central figure of all human history.I know He lives. I know He was born. I know He has been resurrected. I know He is our Savior and Redeemer. And with you I rejoice in this beautiful… pic.twitter.com/YDQpIzgk0n\n\n— Neil L. Andersen (@AndersenNeilL) December 17, 2023\n\n\n\n\nDid He come?\nNothing Elder Anderson shares here is new. It’s rather repetitive. Same old story. Same-old old-guy telling me to believe in a Jesus.\nIf you consider the most currently interesting things in this world, Artificial Intelligence, some may tell you that there exists a future utopia where we’re supported by infinitely intelligent digital agents who can solve complex world problems and bring us to…utopia.\nOk. We’ve also heard this before. The Tower of Babel tried to approximate heaven, and now humanity is trying to approximate Godlike intelligence.\nThe void of Christ in one’s life produces a desire to create Christ. A savior. Something to literally save us from our woes.\n\n\ni think AI is going to be the greatest force for economic empowerment and a lot of people getting rich we have ever seen\n\n— Sam Altman (@sama) February 13, 2023\n\n\nLet’s be real. I’m a tech optimist. I say bring it on. But I’m also a Christian. In my view, the most important question isn’t “what will AGI do to/for humanity?” Instead, the most important question is “was Jesus Christ the promised messiah?”\nIf Christ came, and was indeed the prophesied Christ, then all of life’s ultimate questions are already answered.\n\nWhy are we here? To die and be resurrected to live again.\nWhere did we come from? The same place He came from: the presence of God.\nWho are we? We are sons and daughters of God worth saving.\n\nYes, the questions are already answered: it doesn’t matter what happens with AI. Because the end is known: the Lord will come again.\nLet us treasure this time to reflect on a Savior who already came. Who lives. Who loves us beyond mortal (or AI) comprehension.\n\n\nWhat can we do?\nLet’s take a moment, perhaps an hour, to consider the miracle of Christmas:\n\n\nAs Christmas approaches, I realize that some may have concerns and perhaps some anxiety about the future. There may be a lot of “noise” in your life: more or less constant engagement online without downtime, without time to be quiet and reflect and think, without time to look… pic.twitter.com/4rqS2F6No9\n\n— D. Todd Christofferson (@ChristoffDTodd) December 15, 2023\n\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Link to the 1912 publication, Twas the Night Before Christmas\noriginal poem was 1820’s\nSt Nick was 4th century\nDutch brought the name “Sinterklass” to the U.S. in the 1600’s.\nSt Nick becomes red by 1931, thanks to Coke\n\n\n\n\n\n\n\nThe OG St. Nick\n\n\nEverything below is produced by GPT:\n“How long has the concept of Santa been around? Where did”st nick” come from?”\n\nThe concept of Santa Claus has evolved over many centuries, originating from the historical figure of Saint Nicholas. Saint Nicholas was a 4th-century Christian bishop in Myra, which is now part of modern-day Turkey. He was known for his generosity and for performing miracles, which earned him a reputation as a protector of children and sailors.\n\n\nThe name “Santa Claus” evolved from the Dutch “Sinterklaas,” a short form of Sint Nikolaas (Saint Nicholas). Dutch immigrants brought the traditions of Sinterklaas to America in the 17th and 18th centuries. In these traditions, Sinterklaas was depicted as a serious, religious figure who rewarded good children and punished the naughty ones.\n\n\nDuring the 19th century, especially in America, the image of Santa Claus as we know it today began to take shape. The 1823 poem “A Visit from St. Nicholas” (commonly known as “Twas the Night Before Christmas”) played a crucial role in shaping the modern character of Santa Claus, including his physical appearance and the tradition of delivering gifts to children.\n\n\nLater, in the 19th and 20th centuries, Santa’s image was further popularized through various media, including advertisements, films, and songs, which contributed to the jovial, red-suited, and reindeer-driving character recognized globally today. This character represents a blend of the historical Saint Nicholas, folk traditions from various cultures, and modern commercial and media influences.\n\nHow did St Nick become red?\n\nThe transformation of Saint Nicholas into the red-suited Santa Claus is a gradual evolution influenced by cultural, artistic, and commercial factors. The image of Santa Claus wearing red clothes began to become popular in the 19th century, but it was solidified in the 20th century, particularly through advertising.\n\n\n\n19th Century Literature and Art: In the 19th century, Santa Claus was often depicted in various colors of robes, including blue, green, and brown. However, by the latter half of the 19th century, the image of Santa in a red suit began to appear more frequently in literature and illustrations. Thomas Nast, a prominent American cartoonist, played a significant role in popularizing this image. In his illustrations for Harper’s Weekly starting in the 1860s, Nast depicted Santa Claus in a variety of outfits, including the now-familiar red suit.\n\n\n\n\nCoca-Cola Advertising in the 20th Century: The standardized image of Santa Claus in a red suit with white fur trim was popularized and became iconic largely due to Coca-Cola advertising campaigns. In 1931, Coca-Cola commissioned illustrator Haddon Sundblom to create a Santa Claus for their Christmas advertising campaign. Sundblom’s Santa was jolly, rotund, and dressed in a red suit with white fur trim, which matched the company’s color scheme. These advertisements were widely circulated and became immensely popular, cementing the image of Santa Claus in a red suit in the public’s imagination.\n\n\n\nWhile the red-suited Santa Claus is now ubiquitous, it’s important to note that the evolution of his image was influenced by a blend of historical, cultural, and commercial factors over several decades.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "christianity/unknown-god/index.html",
- "href": "christianity/unknown-god/index.html",
- "title": "An Unknown God",
+ "objectID": "posts/2023-12-16-humanity-and-ai/index.html",
+ "href": "posts/2023-12-16-humanity-and-ai/index.html",
+ "title": "What is humanity",
"section": "",
- "text": "This post speaks to how I manage spiritual doubts, such as whether God exists. It got me thinking after reading a few religious posts by Rob Hybdman, someone whose professional (data science) work I greatly admire.\nHere’s a summary of some points I try to make here:\n\nHe doesn’t try to convince us to believe Him without a doubt, He asks us to follow him\nPeople who saw Christ didn’t even recognize Him as the Savior\nChristianity is hard to grasp and won’t make total sense when you think deeply about it.\nOur job is to keep learning, and ask Him to teach us about His nature\n\n\n\n\nI think too many people grow up thinking of God in black/white terms and when one, linking piece of the puzzle doesn’t fit or is missing, some throw Him out all together. Here’s my take on why people misrepresent God, and how this can lead to personal doubts of faith in Him and His Son.\n\n\n\nChrist asking his disciples, “Whom say he that I am?”\n\n\n\nAn unknown God\nA certain philosopher named Korihor once reasoned that people should stop paying tithing or following restrictive religious commandments because they’re wasting their time. Here’s his thinking:\n\nYea, they durst not make use of that which is their own lest they should offend their priests, who do yoke them according to their desires, and have brought them to believe, ==by their traditions and their dreams and their whims and their visions== and their ==pretended mysteries==, that they should, if they did not do according to their words, ==offend some unknown being, who they say is God==—a being who never has been seen or known, who never was nor ever will be. - Alma 30:28\n\nI love this because the argument of whether God exists or if Christ was the Savior is nothing new. (And many people I highly regard and esteem are tackling this “new” issue head on, as many in the world are questioning their faith.) But a lot of the same arguments that have existed continue to exist: why believe?\n…It’s not like anyone’s ever seen God anyway.\nLet’s assume for a minute that they have seen God. They saw His Son, His mortal example, walk the earth (John 5:19).\nWhile with His disciples, Christ himself took a moment to reflect on who people thought he was, and whether people actually recognized Him for who He is:\n\nWhen Jesus came into the coasts of Cæsarea Philippi, he asked his disciples, saying, ==Whom do men say that I the Son of man am?== And they said, Some say that thou art John the Baptist: some, Elias; and others, Jeremias, or one of the prophets. Matthews 16:13-14\n\nPeople had no idea who Christ was even though I’d bet either He or His disciples them His name many times while He was healing them.\nPeople are funny like that.\nConsider your own life. Have you ever misinterpreted a situation, even as it’s being explained to you?\nLet’s consider a magician who does a magic trick for you. She doesn’t show you how it’s done, but she does explain it to you with only words. Since you don’t specifically see how the magic trick is done, perhaps your mind wanders and you doubt what really happened.\nDo you trust the Magician? When He tells you who He is, after healing you, do you believe Him? When an unexplainable coincidence happens in your life, what do you chalk it up to?\n\n“Nah, that’s so highly improbable it must just be luck”… (whatever luck is…)\n\nI can’t imagine how it must have felt for people around Christ to be healed by rubbing mud on their eyes or washing in a river a few times or, or, … I can only honestly believe these people felt absolute wonder to the point of mental disassociation. How else do you mentally cope with being a paralytic your whole life and then you’re able to walk?\nIt’s entirely reasonable, then, that they just figured He must be like the other miracle workers they knew - Elias, John the Baptist, etc. They try to frame Him in their believable mental reference, i.e., “luck”.\n\n\nA strange God\nPerhaps though we can relate to when Paul, a lifelong Non-Christian and Christian persecuter, preached to his fellow Gentiles:\n\n16 Now while Paul waited for them at Athens, his spirit was stirred in him, when he saw the city wholly given to idolatry. 17 Therefore ==disputed he in the synagogue with the Jews==, and with the devout persons, and in the market daily with them that met with him. 18 ==Then certain philosophers== of the Epicureans, and of the Stoicks, encountered him. And some said, ==What will this babbler say?== other some, He seemeth to be a setter forth of strange gods: because he preached unto them Jesus, and the resurrection. Acts 17:16-18\n\nPaul got them. He was once like them: a gentile. An unbeliever. A disbeliever. But he eventually was converted to knowing Christ, and, as such, appeared to seem like a babbler when trying to talk to people who once were like him. They considered him and his teachings strange:\n\n19 And they took him, and brought him unto Areopagus, saying, ==May we know what this new doctrine==, whereof thou speakest, is? 20 For ==thou bringest certain strange things to our ears==: we would know therefore what these things mean. Acts 16:19-20\n\nMeeting Christ can be strange.\nUnderstanding him, so too.\n\n\nPutting a box on God\nConsider the question, “why does God let good people suffer?”\n(Other important questions like, “why did God encourage Abraham/Isaac/Israel to be polygamists but Adam not?”, or “why did God restrict the entire earth from receiving his Gospel until after Christ died, commanding His people not to marry or mingle with them but then change his tone ~33 AD?”, are worth similar consideration.)\nIt’s a false dichotomy: if God exists and loves us, He surely wouldn’t let us suffer. If suffering exists, especially of good1 people, God must not then exist.\n1 What is Good without God?2 The world is filled with unjust, seemingly unnecessary suffering. Consider innocent children starving across the world, being bombed in Gaza/raped in Israel, bombed in Ukraine, Yemen, gassed in Germany in the Holocaust, starved in Stalin’s camps, nuked by the United States in Japan, etc, etc, etc. (I sincerely don’t write this easily, nor to make light of the suffering.) all of the suffering is horrible. But it doesn’t for a moment mean God doesn’t exist. It actually affirms His existence, because if evil exists the opposite of evil must too.3 I don’t mean that in the Marxist political sense, because who’s to declare who’s suffering at any given moment? Is an affluent child born with cancer not a sufferer despite them not being poor?Consider all the suffering of innocent children2. The reality of human nature is sufferers vs non sufferers3. Any suffering is painful to bear, and nobody better understands this than Christ himself (D&C 19:16-19).\nThe point is, we can falsely equivocate that because God loves us and He had the power to stop suffering, He should stop it.\nDo you do this with any other relationship? Consider the hypothetical:\n\nFather Joe loves his children. But Joe let his son Bob get hit with a toy by little sister Susie. Joe let Susie suffer. Therefore, Joe acted in an evil and therefore does not exist.\n\nThis line of thinking is absurd to anyone with a brain, let alone to anyone who is a parent. They understand that sufferig is part of life, can be good for teaching, and that it doesn’t mean the parent doesn’t exist. Yet sometimes we wonder if God exists because we’re suffering.\n\n\nConsider alternative viewpoints\nTo be comfortable with unanswered questions from a yet unknown God, we should consider how the question is already answered.\nFor example, did Hod stop Christ from suffering? No. Did Christ stop himself from suffering on the cross when He had the power to turn water to wine for the thirsty, heal the suffering swordsman’s ear, or even raise Himself from the dead?\nI’m trying to rally around the point: we set up for ourselves false philosophical arguments about who or what God is and should be like when there’s a lot of nuance to be considered.\nIn my view, He’s more of a Magician who “explains but doesn’t show” than one who tries “to convince us about every little thing”. Because it’s quite human to be like “oh, yea that magic trick isn’t cool anymore” once you understand it. And His goal is to invite us to believe His son so that we’ll follow Him. His goal isn’t for us to know everything, though he is willing to teach us a lot. Remember: “ask and ye shall receive”.\n\nAnd now come, saith the Lord, by the Spirit…and let us reason together, ==that ye may understand;== D&C 50:10\n\nHe just didn’t say when He’d tell us.\n\n\nWhy he explains but doesn’t try to convince\nWhy do I believe that He likes to explain but not convince? A few ideas:\n\nHe taught people they would get to heaven by eating His flesh (sacrament), and people misunderstood and thought this was a “hard” teaching. They walked no more with him (John 6), but he didn’t run them down to try to bring them back. Why didn’t he?4\nHe was chained by the Romans, spat on, crowned with thorns, and whipped. But instead of showing them signs and wonders He let them abuse Him. Then He forgave them after they nailed Him to the cross multiple times and speared Him in His side and gave Him vinegar to drink (Psalm 69, John 19).\nHe knew He couldn’t convince them. He once healed a withered man’s hands and the Pharisees wanted to destroy Him because he did so on the sabbath (Matthew 12:9-14)5. How too are we like the Pharisees where we say “God shouldn’t be like X!!!” Shouldn’t we instead let God reveal Himself?\n\n4 Consider this in a humorous way: Why didn’t He say “oh sorry!! Please come back! I’ve got a quota of disciples to reach and if I don’t have you with me…””5 I love how instead of engaging with the Pharisees, He “withdrew” from them and “healed them all”, kind of as a divine rebel just saying, “you’re mad about me healing one guy on the Sabbath? Fine, I’ll heal everyone.”These three clues, among many others, give me perspective on why He’s okay being misunderstood by us.\n\n\nA God of purpose, children of purpose and promise\nWhy does He want us to live by faith? Lots of reasons. But there’s only one I’ll speak to: He’s a God of purpose:\n\n37 And the Lord God spake unto Moses, saying: The heavens, they are many, and they cannot be numbered unto man; but they are numbered unto me, for they are mine. 38 And as one earth shall pass away, and the heavens thereof even so shall another come; and there is no end to my works, neither to my words. 39 For behold, ==this is my work and my glory—to bring to pass the immortality and eternal life of man==. - Moses 1:37-39\n\nHis goal is to prepare an environment to teach and challenge us. Our goal should be to learn who, how and why.\nOur purpose should not be to put God in a box or imagine Him black and white. It should be to learn who He is, so He’a no longer an unknown God.\n\nAnd this is life eternal, that they might know thee the only true God, and Jesus Christ, whom thou hast sent. - John 17:3\n\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "AI and humanity\n\n\nI code in cursor. I use ChatGPT when that fails. I see Pika and Runway for text to video. I see midjourney for photos. Other things that just use GPT to make blog posts.\nWill AGI even solve the problem of getting everything to connect to everything?\nDo we want a stochastic assistant to have access to my bank account, text messages, medical records, social media accounts, etc?\nWill LLMs be the space between human communication? If we’re all using bots to generate material to send to each other, and all use bots to summarize that, then where will the real human connection exist?\nWe all want a world where our problems go away, but if we don’t have problems then are we human? To be naturally human is to be discontent and desire more, and it takes the wise to overcome these impulses. But the masses are not and never have been wise. Only the individuals.\nThe future is bright. But where does humanity fit into all of it?\nWhen problems are too easy to solve with AI, humanity gets to focus even more on the really hard problems: agreeing on how to get along.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "christianity/christmas-empty-tomb/index.html",
- "href": "christianity/christmas-empty-tomb/index.html",
- "title": "Post Christmas blues? Remember the empty tombs",
+ "objectID": "posts/2022-06-20-blogging-with-quarto/index.html",
+ "href": "posts/2022-06-20-blogging-with-quarto/index.html",
+ "title": "Blogging with Quarto, and why I don’t use Medium anymore",
"section": "",
- "text": "An empty spot in the room\nJust cleaned up the Christmas tree and all its ornaments and needles. The glittering, shimmering, and sparkling lights that have warmed us for the last month are officially gone. Packed up. Shipped out. Sent home.\nMy child remarked, “it feels so empty without the tree!” I couldn’t help but agree.\nThe thought dawned on me, perhaps this is what the disciples felt once His light left the earth. Some accounts declare there was thundering and lightning and three days of darkness that covered the earth after He gave His life.\n\n\n\nA dead Christmas tree reminds us of a Christ who died for us on a tree, the cross.\n\n\n\n\nAn empty room, an empty tomb\nBut just as His easily preventable departure left friends and parents weeping (because they didn’t fully understand what He was dying for, (Matthew 16:21-23), perhaps we can remember a little of that sadness that they felt as we clean up the tree. The child was born. He was born to break the bonds of death. He died, some describing He died on a tree. And now, my pine tree representing Christ(mas) is also dead.\n\nWho his own self bare our sins in his own body on the tree, that we, being dead to sins, should live unto righteousness: by whose stripes ye were healed. (1 Peter 2:24)\n\nPerhaps it’s also symbolic we cut down the tree to put it in our homes. It has an expiration date, just like Christ’s life. He knew it was time (John 13:26-27), even if nobody else could comprehend.\nBut remember what happened after? He resurrected. He brought His light back to His friends.\nPerhaps New Year’s lights can remind us of His return. And perhaps a New Year’s resolution can remind us that He invited His disciples to change their ways and go and preach to the gentiles (people previously not allowed to have His gospel and His gift of baptism and the Holy Ghost), to share the gift of Christ(mas) with everyone.\nThe death of Christ marked the fulfillment of prophecy and the fulfillment of the Law of Moses, which meant that the Jews and the house of Israel were no longer the only ones chosen by God but now everybody could receive the gospel (Galatians 3:29).\nLet’s take this new year to remember all are loved. All are remembered. All are invited to happiness and peace through Christ.\n\n\n\nDead tree? Remember the living Christ!\n\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Been here before, eh?\nI probably shouldn’t write this post considering I’ve also written posts “blogging with Jekyll”, “Blogdown”, and “distill” before. I’m flaky, what can I say. I’ve changed my blogging framework 7 times in 5 years (I started with blogger.com and Wordpress before Jekyll, and even used Medium for 25 posts.)\nBut it’s not my fault. There are too many static site generators and it seems that Rstudio is as flaky as me. Meaning, they built rmarkdown, created hugodown, blowdown, bookdown, and distill. Hugodown is a lightweight version of blogdown, and I never figured out why it exists. Then there’s workflowr and so many other great, random products out there.\nWhy quarto for me? It seems to have all the stuff that distill has, and then some. And given its design and collaboration with Wes McKinney (pandas guy) in building quarto, and given that Wes wrote his latest book with it, I think this will be here to stay for at least 5 years.\n\n\nWhy do I avoid Medium\nWhy don’t I blog with Medium? I wrote 25 posts there over 9 months and loved it. It makes writing and publishing incredibly easy, which is why I did it. With distill/blogdown the limiting factor was I’d have to be at my computer to get a post in. I love writing on my phone - I have over 2000 journal entries with Day One over seven years, several of which are “draft” posts for a blog that won’t see the light of day since I don’t care about them after I’ve written it. I want to be able to blog frequently, a short little blurb to keep up my writing, and medium enabled this.\nBut 1) I didn’t get many views (not that I care) and 2) my writing is stuck on their platform (confirmed to me this week as I’ve tried several old, broken tools that attempt to export posts out of medium).\nI got 50x the views on my blogdown site just from organic google searching. From reading other people’s posts about leaving Medium, it’s seems that most traffic to Medium posts comes from outside of Medium.\nSo while I’m certainly not into blogging for the money, I wouldn’t mind getting the content to people who it can help. And if medium didn’t enable this, what is its purpose?\nSo I’m back and finally decided that I’m going to own my own platform. Seth Godin wrote a blog about this topic once. Why be on Medium, substack, Facebook, etc or any other up an coming platform? We’ll they’re all skewed toward the 1% of people who make money on the platform. Everyone else is wasting their time. So might as well own your content. And better to focus on writing than waste time catching up on every platform.\nAlso, it’s fun! There’s a fuzzy feeling to blogging.\n\n\nFuture of quarto?\nI’m speculating here, but Quarto can be used withR Studio, Jupiter or VS Code. The last editor seems excitingly suspicious.\nPerhaps R Studio sees the writing on the wall that VS Code is taking over the IDE world and perhaps taking over the long term future of R Studio IDE. I personally don’t see why anyone would pay for R Studio when VS Code is free and has a better interface. And while R Studio has a few better integrations for R at the moment, it won’t once enough people see the light for VS Code and turn their contributions there. I’ve been using R on VSCode for two years now and it rivals R Studio in 95% of the feature set, but 100% has better window management.\nSo Quarto seems like a great opportunity for R Studio to begin monetizing on a publishing platform like R Studio connect. Platforms like Notion and Confluence are great for everyone except data scientists trying to share reproducible research. But a quarto server seems like a great way to finally bring in Python data scientists into the markdown-flavored documentation world. (Man, I personally can’t use Jupyter/Google colab notebooks when I can use R Markdown!) I’d heartedly welcome this prospect, though I have no idea what their plans for Quarto really are.\nTill next blog post about what blogging tool I’m going to use…\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "index.html",
- "href": "index.html",
- "title": "Recent Posts",
+ "objectID": "posts/2022-12-06-more-new-businesses-than-ever/index.html",
+ "href": "posts/2022-12-06-more-new-businesses-than-ever/index.html",
+ "title": "Models as a Service and the Future of Data Science",
"section": "",
- "text": "“All blogs are wrong, but some are useful.”\n\n\n\n\n\n\n \n \n \n Order By\n Default\n \n Title\n \n \n Date - Oldest\n \n \n Date - Newest\n \n \n \n\n\n\n\n\n\n\n\n\n\nPost Christmas blues? Remember the empty tombs\n\n\n\n\n\n\nchristianity\n\n\n\nAn empty room and an empty tomb. There is meaningful symbolism in taking down the tree. \n\n\n\n\n\nWed Dec 27, 2023\n\n\n3 min\n\n\n\n\n\n\n\n\n\n\n\n\nAn Unknown God\n\n\n\n\n\n\nchristianity\n\n\ntheology\n\n\nfaith\n\n\ndoubts\n\n\n\nGod is indeed unknown, but that doesn’t mean He doesn’t exist. And learning His ways requires seeing all sides, being intellectually and spiritually honest, and open. \n\n\n\n\n\nWed Dec 27, 2023\n\n\n10 min\n\n\n\n\n\n\n\n\n\n\n\n\nWriting is truly enlivening\n\n\n\n\n\n\ncommunication\n\n\nwriting\n\n\n\nWrite in public. Write daily. Write about nothing. \n\n\n\n\n\nTue Dec 26, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nWrite despite AI\n\n\n\n\n\n\nai\n\n\nhumanity\n\n\n\nI’m fully aware an AI is listening and all that I write just trains it. Perhaps this is a perversion of my work (if I made money off it). Why write if everything I write just trains an AI agent to do their job better? \n\n\n\n\n\nTue Dec 26, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nThings I’d like in a blog\n\n\n\n\n\n\nproductivity\n\n\nquarto\n\n\n\nWhat quarto is great at and what I’d like out of it. \n\n\n\n\n\nTue Dec 26, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nI’m grateful for blankets\n\n\n\n\n\n\ngratitude\n\n\n\nMy grandma made blankets cool \n\n\n\n\n\nTue Dec 26, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nHow to blog from your phone using quarto (or hugo/jekyll/etc) using Working Copy and Obsidian\n\n\n\n\n\n\nproductivity\n\n\ntech\n\n\ntutorial\n\n\n\nWorking copy is $20, obsidian is free (and optional). I’ll show you how to set this up (iPhone only I think?) \n\n\n\n\n\nTue Dec 26, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nI’m grateful for doctors\n\n\n\n\n\n\ngratitude\n\n\n\nDoctors of the body. Doctors of the mind. Doctors of the soul…Any engaged in the pursuit of healing. \n\n\n\n\n\nMon Dec 25, 2023\n\n\n3 min\n\n\n\n\n\n\n\n\n\n\n\n\nAI Robots are mastering tasks faster than humans\n\n\n\n\n\n\nai\n\n\ntech\n\n\nfuturism\n\n\n\nFrom walking to labyrinth, the future looks bright for these machines. The key is the objective functions, time, and training data. \n\n\n\n\n\nMon Dec 25, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nTwas the Night before Christmas\n\n\n\n\n\n\nchristmas\n\n\nask gpt\n\n\n\nThe origin of the story, and a link to the 1912 publication. \n\n\n\n\n\nSun Dec 24, 2023\n\n\n3 min\n\n\n\n\n\n\n\n\n\n\n\n\nThe Christ Child\n\n\n\n\n\n\nchristianity\n\n\n\nThe birth of Jesus Christ teach us what we can do to have greater peace in this life \n\n\n\n\n\nSun Dec 24, 2023\n\n\n4 min\n\n\n\n\n\n\n\n\n\n\n\n\nLight\n\n\n\n\n\n\ngratitude\n\n\n\nI set up 72 lights today. \n\n\n\n\n\nSun Dec 24, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nHis is the light that we should share\n\n\n\n\n\n\nchristianity\n\n\n\nWho is the source of light? \n\n\n\n\n\nSun Dec 24, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nThe Best of Edgar Allen Poe\n\n\n\n\n\n\nbook review\n\n\n\nYou think me mad? \n\n\n\n\n\nSat Dec 23, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nDune\n\n\n\n\n\n\nbook review\n\n\n\nThe plans and risks of generations \n\n\n\n\n\nSat Dec 23, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nThe Killers of the Flower Moon\n\n\n\n\n\n\nbook review\n\n\n\n\n\n\n\n\n\nFri Dec 22, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nThe Allegory of the Dodo\n\n\n\n\n\n\nproductivity\n\n\nbusiness\n\n\nkids books\n\n\n\nWhen in having solved a problem we inadvertently create problems for ourselves \n\n\n\n\n\nThu Dec 21, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nGitHub copilot levels up\n\n\n\n\n\n\nai\n\n\nproductivity\n\n\ntech\n\n\n\nWhat will it do next? \n\n\n\n\n\nThu Dec 21, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nThe Little Red Hen\n\n\n\n\n\n\nbook review\n\n\nstartups\n\n\n\nPeople love to enjoy the fruits, but who will bake the bread? \n\n\n\n\n\nWed Dec 20, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nHow to encourage your kids\n\n\n\n\n\n\nparenting\n\n\n\nUnexpected advice on how to be a great parent when fostering talent \n\n\n\n\n\nTue Dec 19, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\n101 Essays that will Change the Way You Think, by Brianna Wiest\n\n\n\n\n\n\nbook review\n\n\nphilosophy\n\n\nmindfulness\n\n\n\nWe all started as strangers. It’s a choice who remains one. \n\n\n\n\n\nTue Dec 19, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nHitchhiker’s Guide to the Galaxy\n\n\n\n\n\n\nbook review\n\n\nstorytelling\n\n\n\nHow did it take me so long to finally read this? \n\n\n\n\n\nMon Dec 18, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nChristmas celebrates a Living Christ\n\n\n\n\n\n\nchristianity\n\n\nai\n\n\n\nChrist is the way, the light and life of the world. Always has been. Always will be. \n\n\n\n\n\nSun Dec 17, 2023\n\n\n3 min\n\n\n\n\n\n\n\n\n\n\n\n\nWhat is humanity\n\n\n\n\n\n\nai\n\n\n\nHumanity seeks to resolve problems. What happens when they’re too easily solved? \n\n\n\n\n\nSat Dec 16, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nPrompt techniques\n\n\n\n\n\n\nai\n\n\n\nOpen AI strategies \n\n\n\n\n\nFri Dec 15, 2023\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nLLMs can find a needle in the haystack\n\n\n\n\n\n\nai\n\n\n\nGPT outperforms Claude. \n\n\n\n\n\nFri Dec 15, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nThree Little Pigs\n\n\n\n\n\n\nbook review\n\n\nchristianity\n\n\n\nA story with a terrible but incredibly relatable morale \n\n\n\n\n\nWed Dec 13, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nLearn by doing\n\n\n\n\n\n\ndata science\n\n\ntech\n\n\n\nYou don’t need a PhD. Or a masters. You just need to hustle \n\n\n\n\n\nWed Dec 13, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nScientific Freedom\n\n\n\n\n\n\nbook review\n\n\n\nInnovation comes by varied incentives: intellectual curiosity and economic incentive \n\n\n\n\n\nTue Dec 12, 2023\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nFuture of Coding\n\n\n\n\n\n\ndata science\n\n\n\nAs our jobs are automatable, we need to discover what’s not \n\n\n\n\n\nTue Dec 12, 2023\n\n\n4 min\n\n\n\n\n\n\n\n\n\n\n\n\nHow do you want to be remembered?\n\n\n\n\n\n\nleadership\n\n\nservice\n\n\n\nIt’s a question as old as time, but misses the mark and is distracting from the more important question. \n\n\n\n\n\nFri Dec 23, 2022\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nBuilding a complex R Shiny Dashboard Using ChatGPT\n\n\n\n\n\n\ndata science\n\n\nr\n\n\nshiny\n\n\nfuturism\n\n\nchatgpt\n\n\n\nChatGPT can create amazing boilerplate code and can point you in the right direction, but it cannot debug data problems easily (yet). It can read in data from the internet, make time-series forecasts, and visualize the data (with a little assistance). \n\n\n\n\n\nThu Dec 15, 2022\n\n\n8 min\n\n\n\n\n\n\n\n\n\n\n\n\nModels as a Service and the Future of Data Science\n\n\n\n\n\n\ntech\n\n\nstartups\n\n\nbusiness\n\n\ndata science\n\n\n\nData Scientists shouldn’t be scared their jobs will disappear. They should be excited that they’ll be joining companies that automate what they’re currently doing and offer it as a service. \n\n\n\n\n\nTue Dec 6, 2022\n\n\n4 min\n\n\n\n\n\n\n\n\n\n\n\n\nWriting a Tech Resume\n\n\n\n\n\n\ncommunication\n\n\n\nHow to write a resume to get into a FAANG company. \n\n\n\n\n\nFri Nov 18, 2022\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nResume Substance over Style\n\n\n\n\n\n\ncommunication\n\n\n\nThe content of your resume is more important than the design. \n\n\n\n\n\nFri Nov 18, 2022\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nWhat is a model?\n\n\n\n\n\n\ndata science\n\n\n\nPutting the mysterious in context. \n\n\n\n\n\nSat Nov 12, 2022\n\n\n5 min\n\n\n\n\n\n\n\n\n\n\n\n\nDe-sciencing Data Science and Talking Like a Normal Person\n\n\n\n\n\n\ndata science\n\n\ncommunication\n\n\nprinciples\n\n\n\nHow can data scientists bring their technical knowledge to a non-technical audience? Here are my lessons learned from seven years in the data trenches. \n\n\n\n\n\nWed Jul 20, 2022\n\n\n6 min\n\n\n\n\n\n\n\n\n\n\n\n\nWhen data collection goes too far\n\n\n\n\n\n\nproductivity\n\n\nfuturism\n\n\ntech\n\n\nsurveillance\n\n\n\nIt’s important for us to track certain things about ourselves. But this is clearly not exactly healthy IMO. \n\n\n\n\n\nSun Jul 3, 2022\n\n\n4 min\n\n\n\n\n\n\n\n\n\n\n\n\nPomodoro Principles\n\n\n\n\n\n\nproductivity\n\n\nticktick\n\n\n\nThe art of focus through 25min work blocks. \n\n\n\n\n\nTue Jun 28, 2022\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nIncrease Focus through Writing Down Everything\n\n\n\n\n\n\nproductivity\n\n\nmusings\n\n\nticktick\n\n\n\nTo be more present, get it all down and into a system. \n\n\n\n\n\nTue Jun 28, 2022\n\n\n3 min\n\n\n\n\n\n\n\n\n\n\n\n\nHow I use TickTick\n\n\n\n\n\n\nproductivity\n\n\n\nAn overview of how I apply the principles of “Getting Things Done”, or GTD. \n\n\n\n\n\nSat Jun 25, 2022\n\n\n7 min\n\n\n\n\n\n\n\n\n\n\n\n\nGetting Things Done\n\n\n\n\n\n\nbook review\n\n\nproductivity\n\n\n\nA powerful system to be more present. \n\n\n\n\n\nFri Jun 24, 2022\n\n\n6 min\n\n\n\n\n\n\n\n\n\n\n\n\nWhat is the best todo list manager?\n\n\n\n\n\n\nproductivity\n\n\napp review\n\n\n\nI’ve reviewed 20+ apps. I think I’m done reviewing them. \n\n\n\n\n\nThu Jun 23, 2022\n\n\n11 min\n\n\n\n\n\n\n\n\n\n\n\n\nNew ideas are distractingly exciting\n\n\n\n\n\n\nproductivity\n\n\npriorities\n\n\nideation\n\n\n\nBe careful of recency bias with your ideas. \n\n\n\n\n\nThu Jun 23, 2022\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nDownstream Impact\n\n\n\n\n\n\ndata science\n\n\nexperimentation\n\n\n\nWhen you do an online experiment, you’re changing the future forever. \n\n\n\n\n\nWed Jun 22, 2022\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nThe Power of Party Pay by Visible\n\n\n\n\n\n\nstartups\n\n\nstrategy\n\n\n\nVisible wireless, owned by Verizon, offers users $25 off per month if they join a meaningless party. What’s the effect? \n\n\n\n\n\nTue Jun 21, 2022\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nProducts for dads\n\n\n\n\n\n\nproductivity\n\n\nfatherhood\n\n\n\nWhat products am I still using 5 years later? \n\n\n\n\n\nTue Jun 21, 2022\n\n\n6 min\n\n\n\n\n\n\n\n\n\n\n\n\nBlogging with Quarto, and why I don’t use Medium anymore\n\n\n\n\n\n\ndata science\n\n\n\nI’m trying out quarto and I like it. \n\n\n\n\n\nMon Jun 20, 2022\n\n\n4 min\n\n\n\n\n\n\n\n\n\n\n\n\nThe Author’s Angle Matters\n\n\n\n\n\n\ncommunication\n\n\nwriting\n\n\n\n\n\n\n\n\n\nSat Jul 17, 2021\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nProductivity notes\n\n\n\n\n\n\ndata science\n\n\nproductivity\n\n\n\nRunning list of productivity ideas. \n\n\n\n\n\nSat Dec 19, 2020\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nHow to mock the value of an ML solution\n\n\n\n\n\n\ndata science\n\n\ninnovation\n\n\n\nSell your solution before training your model. \n\n\n\n\n\nWed Dec 16, 2020\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nGet rid of the todo list. Calendar everything.\n\n\n\n\n\n\nproductivity\n\n\n\nIf you’re not willing to block time for it, does it event matter? \n\n\n\n\n\nTue Dec 15, 2020\n\n\n1 min\n\n\n\n\n\n\n\n\n\n\n\n\nData Beats Opinions\n\n\n\n\n\n\ndata science\n\n\ndecisions\n\n\n\nOpinions from the customer shape the product. Their behavior (captured as data) speaks louder than their words. \n\n\n\n\n\nTue Dec 15, 2020\n\n\n2 min\n\n\n\n\n\n\n\n\n\n\n\n\nSemantic Versioning for Data Science Models\n\n\n\n\n\n\ndata science\n\n\nmachine learning\n\n\n\nMeaningful versioning for data science models and machine learning pipelines. \n\n\n\n\n\nMon Jul 2, 2018\n\n\n5 min\n\n\n\n\n\n\n\n\n\n\n\n\nBuying a used car the data science way: Part 2\n\n\n\n\n\n\nwebscraping\n\n\nr\n\n\nregression\n\n\nanalysis\n\n\npricing\n\n\nbuying things\n\n\n\nHow I analyze used car data to find under-valued cars, and why none of my analysis matters.\n\n\n\n\n\nSun Feb 19, 2017\n\n\n14 min\n\n\n\n\n\n\n\n\n\n\n\n\nBuying a used car the data science way: Part 1\n\n\n\n\n\n\nwebscraping\n\n\ncars\n\n\npricing\n\n\nr\n\n\n\nHow I scrape used car data.\n\n\n\n\n\nSat Feb 18, 2017\n\n\n4 min\n\n\n\n\n\n\n\n\n\n\n\n\nThe DataViz battle: Plotly vs ggplot2\n\n\n\n\n\n\ndataviz\n\n\nr\n\n\nggplot\n\n\nplotly\n\n\n\nWhat can you do with plotly vs. ggplot2 and how do they compare on a simple chart?\n\n\n\n\n\nFri Feb 10, 2017\n\n\n7 min\n\n\n\n\n\n\n\n\n\n\n\n\nStarting with Jekyll\n\n\n\n\n\n\nproductivity\n\n\nblogging\n\n\njekyll\n\n\n\nHello world.\n\n\n\n\n\nThu Nov 17, 2016\n\n\n5 min\n\n\n\n\n\n\nNo matching items\n\n\n \n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only.",
- "crumbs": [
- "Recent Posts"
- ]
+ "text": "Businesses are Launching Like Crazy\nThere’s never been a better time to start a business. The Census shows there are over 200,000 new business applications per month more than pre pandemic.\n\nWhat’s driving this?\nListening to The Unintended Consequences of Working from Home it dawned on me that there’s a hyper acceleration of innovation happening.\nConsider the rise of AI tools like ChatGPT and StableDiffusion. Chamath Palihapitiya discusses the expansive applications for these services. What was once SaaS will become MaaS (models as a service). There will be models for everything: copy writing, code debugging, converting wireframes into working apps, ad generation, book writing…\nWhile tools like ChatGPT are just in demo mode, it’s wildly apparent they’ll be an impetus for new ideas and reducing the friction of getting to the answer. For example, as a data scientist if I need boilerplate code to clean some data, I can ask ChatGPT for the 80% solution. I’m already spending 30% of my day copying code snippets from StackOverflow and making it work for my problem. ChatGPT will just accelerate that process. 10 years from now if Neuralink can read my thoughts I don’t even need a keyboard to do this…we’ll build things at the speed of thought.\nConsider this for writers and marketers. They’ll use ChatGPT to unblock them, generate ideas, and get started. Innovation across the board will accelerate because the kinetic friction to getting started will reduce to zero. You can then spend your time on the editing and refining.\n\n\nThe Best Time to Start a Business\nI digress. Why is now the best time to build a software business?\n\nCompute is cheap. Gone are the days of having to build a server to prototype and idea.\nBarriers to prototyping are nearly free and scaling with the cloud is nearly infinite. Consider how the Gas App can launch to 100,000 new users per day with just a few developers.\nModeling is now easy. All cloud services have their own AutoML tools to build machine learning models for you.\n\nThe center of all of this is scalable data infrastructure and AI.\n\n\nFuture of Data Science\nThe data scientist of today may still be using sklearn to manually do hyperparameter tuning, but the data scientist of tomorrow will either be exclusively building MaaS B2B products or be the one person in their organization who will leverage all the MaaS tools.\nIf we can ask ChatGPT for complex answers it learned from the internet, why couldn’t something similar be trained on your company’s internal data? If that’s possible, then why couldn’t you just have a Q&A session with your internal ChatGPT? Consider the following example:\n\n“Hey KITT, what are the user segments generating the most revenue on Tuesdays at 5pm?” (Analytics)\n“How many sales do we anticipate from these users next Tuesday?” (Machine Learning)\n“What ads have been most effective for us in driving new sales in the past 7 days?” (Statistics)\n“Okay, KITT, generate 3 variants of that ad for next Tuesday” (StableDiffusion generating a new Ad)\n“I like ads A, B and C. launch an experiment on Facebook/TikTok, run it for three days and dynamically allocate ad spend to the best ad.” (Business integrations, statistics, ML models)\n\nThe data scientist of the future will be the one with deep understanding of these tools and an awareness of their biases. It will be very difficult to know when they’re lying to you. The in-house data scientist will be keen on knowing the biases in the data fed into these algorithms and how to make judgments on where to go next.\nEarly stage companies won’t need data scientists because most of this stuff will be off the shelf and easy for an engineer to integrate. That means that the data scientists will join these MaaS companies to build out optimized products that solve these niche problems.\n\n\nTakeaways\nData Scientists shouldn’t be scared their jobs will disappear. They should be excited that they’ll be joining companies that automate what they’re currently doing and offer it as a service.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "gratitude/2023-12-26-i'm-grateful-for-blankets/index.html",
- "href": "gratitude/2023-12-26-i'm-grateful-for-blankets/index.html",
- "title": "I’m grateful for blankets",
+ "objectID": "posts/2020-12-15-data-beats-opinions/index.html",
+ "href": "posts/2020-12-15-data-beats-opinions/index.html",
+ "title": "Data Beats Opinions",
"section": "",
- "text": "A blanket\n\n\nI’m outside right now enjoying the warmth of a blanket. It’s a special thing. It’s warm. It’s fuzzy. And the only question I have is: is it warm because it’s fuzzy? Or is it fuzzy because it’s warm?\nIn the Hitchhiker’s Guide to the Galaxy the most important item any many could carry was a towel. It’s a multifaceted tool. Perhaps a blanket is the towel’s cousin.\nMy grandma loved to quilt, so me and all my family have hand crafted blankets. Even great grandchildren have blankets long after my grandma deceased because she had a surplus in her basement.\nThere’s something special about being wrapped in something warm, especially if it’s made by someone who loves you. Kinda sends a special hug in a way that a towel just can’t.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "When I co-started a company several years ago, my founders and I would canvass neighborhoods knocking doors. We’d discuss our ideas for a new product one door at a time. After each conversation, the product took a new form. Eventually, the product took shape and we had a successful kickstarter launch.\nSince that time, I’ve thought about how valuable it was to get information straight from the source.\nAlberto Savoia’s new book, “The Right It”, is re-teaching me this principle. Data beats opinions, he says. In another book I’m reading, “Trustworthy Online Controlled Experiments”, the former CEO of Netscape is quoted saying,\n\nIf we have data, let’s go with data. If all we have are opinions, let’s go with mine. - Jim Barksdale\n\nData sourced from your manager, your stakeholder, or your customer are invaluable to help you know if you’re solving their problem. Too often it’s easy to get stuck in the original solution that comes to our mind.\nData from customers directly shape the product. Data sourced at scale from online experiments refine it.\nOpinions from the customer shape the product. Their behavior (captured as data) speaks louder than their words. Frequently, it’s challenging to correlate the two. We’re left to guess why someone actually clicked, bought, churned. It’s a beautiful cycle.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2021-07-17-authors-angle-matters/index.html",
- "href": "posts/2021-07-17-authors-angle-matters/index.html",
- "title": "The Author’s Angle Matters",
+ "objectID": "posts/2017-10-19-used-cars-analysis/used-cars.html",
+ "href": "posts/2017-10-19-used-cars-analysis/used-cars.html",
+ "title": "Buying a used car the data science way: Part 2",
"section": "",
- "text": "Imagine the following story:\n\n“In recent events, a ferocious fox savagely attacked a Turkey, who fortunately was able to escape to safety by pecking out the fox’s eyes.”\n\nSimple story. A bit gruesome. But, what in today’s news cycle is positive? How can we be better readers of the news when it’s targeted at making us angry?\nFoxes and turkeys are highlighted in the news all the time and it’s often the worst of humanity. The goal of many authors is to evict emotion from you so you’ll keep reading. Anger. Fear. Worry. Those keep you hooked.\nWhat’s the author’s motive? Often we have to guess it. Wouldn’t it be nice instead if authors were transparent about why they were writing a story:\n\nAuthor 1: I’m a Turkey conservationist with a passion for protecting turkeys from the rampant fox problem we have. Growing up, a Turkey saved my life when I was drowning in a lake.\nAuthor 2: I’m a narcissist who hates foxes because one time they ate my cat. I really just need therapy, but I can’t fathom talking to anyone about my problems. I want you to be as angry as I am.\nAuthor 3: I’m here to get promoted and if my boss sees that I have a 10% increase in viewership then our company’s bottom line increases. Im an opportunist who will say anything and take any side that will maximize company profits. How do these three perspectives change the storyline?\n\nWouldn’t it be great if we could know the author in addition to what the author is writing about? Every article should have not just “about the author” but, “three people’s perspectives about the author’s intent in writing this that know the author well.” Of course, you’d need to verify the three verifiers, but it’d be a start\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "[Update 2021-11-16] This analysis was originally written on my old blog here. You can find the source code for it here.\nThis is part 2 out of a two-part series on scraping used car data. Check out part 1 to learn how to scrape the data.\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2016-11-07-starting-with-jekyll/starting-with-jekyll.html",
- "href": "posts/2016-11-07-starting-with-jekyll/starting-with-jekyll.html",
- "title": "Starting with Jekyll",
- "section": "",
- "text": "Hello world.\nI wanted to start a blog. I wanted to set it up for free, use a custom URL (and not something.bloggingplatform.com), and be able to both blog and create tutorials. I didn’t mind it possibly being technical1. Enter Jekyll.\n1 Specifically, I was interested in being able to use the command line for my posts. Additionally, I didn’t mind seeing or using a little bit of code.If you want to get your blog in 10 minutes, skip to below.\n\nWhy Blog with Jekyll?\nHere’s why you can and should blog with Jekyll (if you’re a data scientist):\n\nJekyll has beautiful, free, open-source templates.\nGitHub will host your website for free2.\nJekyll is static.\nJekyll can be simple to set up3.\nI tried Wordpress and blogger.com and got frustrated.\nThe blogging content you create is very portable (easy to take your posts from one blog site to another).\nCreating blogs in Markdown syntax is a dream (even for non-technical folks).\nYou can manage your blog from your terminal\nYou can manage your blog from your phone (if you have a GitHub-editing app)\n\n2 Someone has to host it (i.e., store and display all the files). I started off trying to set up a WordPress blog, but you have to pay another company to host. I didn’t want to bother.3 If you try to set it up by following the instructions on jekyllrb.com, that’s the complicated way. That’s how I originally started, got lost for five hours, and stopped. Six months later, I found a better way.But if you don’t like debugging occasional errors, you probably shouldn’t blog with Jekyll.\nWhat is Jekyll, you ask? Jekyll is an open-source blogging platform. Anyone can write a template and post it, and anyone else can use that template. Perusing Jekyllthemes.org, you can pick out any template you’d like.\n\n\nHow did I get started with Jekyll?\nAs a blogging noob, I’ve discovered there are hundreds of ways to generate a blog (e.g., WordPress, Squarespace, blogger.com, Weebly, etc.). But when my colleague Arthur Lui showed me his blog, I wanted something similar. I first tried Wordpress, then tried blogspot, then gave up and followed Arthur’s example and landed with Jekyll.\n\n\nSet up a blog in 10 minutes!\nI followed this Jekyll tutorial to set up my first blog. I’ll simplify those steps here to get you up and running quickly:\n\nGet a GitHub account. (2 min)\nGo to Jekyllthemes.org. All of these blog themes are free. Find one you like, and Fork the repo (i.e., “repository”). (2-60 min, depending on how long you peruse)\nIn GitHub, rename the newly-forked repo from “REPO name” to “yourgithubusername.github.io”. (1 min)\nNow go to “githubusername.github.io”. Voila, you have a blog.\n\nAnd if you want to start writing posts, go to your “_posts” folder. Then create new files and follow the default templates that the repository provides. (The blogs will need to be written in Markdown syntax.)\nThat was my 10 minute promise. If you feel a bit cheated or lost, go to the well-done tutorial and do all the steps. But the steps I listed are the essential steps to setting up a blog from scratch. So if you want to make a few blogs really quickly, after you figure out the first one the others follow nicely. For example, once I finished the tutorial, I followed those steps above to try two other themes before deciding on the Centrarium theme.\nTo get Centrarium to work, I did have to change some configurations in the “_config.yml”, so there can be some minor debugging depending on the theme you choose. But with a little patience and some Googling, you’ll get it.\n\n\nAdd to Jekyll a custom URL from GoDaddy\nYou’ll notice I have a custom URL and not “bryanwhiting.github.io”. I used GoDaddy to get a domain. (I have no affiliation with GoDaddy.) Here’s how I linked GoDaddy with my GitHub Jekyll blog:\n\nGet a domain.\nIn your Github repo, change (or create) a CNAME file in the main directory. (See my CNAME file for an example). And for the CNAME file contents, just type in the new “www” that you registered through GoDaddy.\nThen, on GoDaddy, go to the section “DNS Management”. Here’s where you tell GoDaddy how to connect your newly purchased “www” with GitHub. Create the following connections:\n[Type, Name, Value, TTL] = [A, @, 192.30.252.153, 600 seconds]\n[Type, Name, Value, TTL] = [A, @, 192.30.252.154, 600 seconds]\n[Type, Name, Value, TTL] = [CNAME, ftp, yourgithubusername.github.io, 600 seconds]\n[Type, Name, Value, TTL] = [CNAME, www, yourgithubusername.github.io, 600 seconds]\n\nIf you have the CNAME file in your GitHub folder, GitHub and GoDaddy will talk and after a few minutes your custom “www” will work. You don’t technically need the two [A] steps, but those connections allow users to type “bryanwhiting.com” instead of “www.bryanwhiting.com”. It can take up to 24 hours before these [A] connections enable, so don’t be worried if the short URL doesn’t work first try. But the full URL should work soon enough.\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "objectID": "posts/2017-10-19-used-cars-analysis/used-cars.html#youre-in-the-market",
+ "href": "posts/2017-10-19-used-cars-analysis/used-cars.html#youre-in-the-market",
+ "title": "Buying a used car the data science way: Part 2",
+ "section": "You’re in the market",
+ "text": "You’re in the market\nSo you want to buy a car, but you don’t know anything about them? Welcome to my life.\nYou show up at the dealer and there’s a sticker on the window. You know the difference between make and model, but you soon learn what a trim is. Some versions come with leather. Some have a sun roof. Some have all wheel drive. Some have 20k in miles, and a similarly priced car in a higher trim is at 40k miles. How do you know what you’re really paying for, and what these features are worth? And how do you know what it’ll be worth when it’s four years older and has an additional 40k miles?\nPretty advanced questions for someone who just learned what a powertrain is.\nIn this analysis, I’ll dive into how you can use data to learn a lot about an industry in a short amount of time. I’ll first dive into knowing what you’re buying (now and later). Then, I’ll dive into how Truecar and others might be finding their ‘deals’."
},
{
- "objectID": "posts/2020-12-15-calendaring-todos/index.html",
- "href": "posts/2020-12-15-calendaring-todos/index.html",
- "title": "Get rid of the todo list. Calendar everything.",
- "section": "",
- "text": "I’ve had less meetings and more unstructured time during WFH. It can be isolating, which can slow down productivity.\nHere’s a tip I read today on how to be productive with unstructured time: put your to-do list directly into a calendar.\nThat way, you’ll always have time for the important things. Remove the things that take up your time and aren’t as important. I like it.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "objectID": "posts/2017-10-19-used-cars-analysis/used-cars.html#what-am-i-paying-for",
+ "href": "posts/2017-10-19-used-cars-analysis/used-cars.html#what-am-i-paying-for",
+ "title": "Buying a used car the data science way: Part 2",
+ "section": "What am I paying for?",
+ "text": "What am I paying for?\nTo do any analysis, you need data. Please see my most recent post if you’re interested in learning how I got the data for this analysis. Suffice it to say, I webscraped it. The data is not my own, but belongs to Truecar.\nI love using linear regression to understand my data on a granular level. In my data set, I was able to extract the make, model, year, mileage, trim, and a flag for whether the car was all wheel drive (awd == 1) or not. Putting this data into a linear regression, we can see some interesting results. First let’s look at the data we have:\n\nlibrary(dplyr)\nload('../../data/used-car/tc-ford-edge.Rda')\n# Get the prices into the appropriate scale and remove some extraneous\ndf$price = df$price*1000\ndf %>% select(-stats, -url, -int, -vin, -v6, -trueprice, -resid) %>% head()\n\n## price year mileage trim awd location ext deal\n## 1 16000 2014 28.477 SE 1 Sterling, VA <NA> 6514\n## 2 13300 2013 64.315 SE 0 Chantilly, VA <NA> 1784\n## 3 13800 2013 72.111 SEL 0 Fairfax, VA Mineral Gray Metallic 3331\n## 4 16000 2014 27.490 SE 0 Manassas, VA Oxford White 5448\n## 5 12000 2013 66.825 SE 0 Clarksville, MD Mineral Gray Metallic NA\n## 6 14900 2014 65.157 SE 0 Manassas, VA Tuxedo Black Metallic 4436\n## mpyr\n## 1 7.12\n## 2 12.86\n## 3 14.42\n## 4 6.87\n## 5 13.37\n## 6 16.29\nNow let’s see the regression results.\n\nmod <- lm(price ~ as.factor(year) + mileage + trim + awd, data = df)\nsummary(mod)\n\n## \n## Call:\n## lm(formula = price ~ as.factor(year) + mileage + trim + awd, \n## data = df)\n## \n## Residuals:\n## Min 1Q Median 3Q Max \n## -4755.7 -854.6 -2.7 877.5 5024.2 \n## \n## Coefficients:\n## Estimate Std. Error t value Pr(>|t|) \n## (Intercept) 22216.400 788.885 28.162 < 2e-16 ***\n## as.factor(year)2013 1267.037 625.804 2.025 0.0442 * \n## as.factor(year)2014 3160.017 645.152 4.898 1.92e-06 ***\n## as.factor(year)2015 5664.470 710.322 7.975 9.56e-14 ***\n## as.factor(year)2016 6166.658 821.469 7.507 1.68e-12 ***\n## as.factor(year)2017 9082.971 878.956 10.334 < 2e-16 ***\n## mileage -91.951 5.625 -16.347 < 2e-16 ***\n## trimSE -4983.725 409.765 -12.162 < 2e-16 ***\n## trimSEL -1822.889 291.400 -6.256 2.17e-09 ***\n## trimSport 3419.270 374.977 9.119 < 2e-16 ***\n## trimTitanium 99.242 568.428 0.175 0.8616 \n## awd 1277.734 222.680 5.738 3.30e-08 ***\n## ---\n## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1\n## \n## Residual standard error: 1450 on 211 degrees of freedom\n## Multiple R-squared: 0.9231, Adjusted R-squared: 0.9191 \n## F-statistic: 230.2 on 11 and 211 DF, p-value: < 2.2e-16"
},
{
- "objectID": "posts/2022-06-23-new-ideas-are-distractingly-exciting/index.html",
- "href": "posts/2022-06-23-new-ideas-are-distractingly-exciting/index.html",
- "title": "New ideas are distractingly exciting",
- "section": "",
- "text": "When a new idea comes to my mind, I want to act on it right away. Perhaps it’s an exciting analysis idea, a new business strategy I want to scope out, or a question to ask a team member.\nBut as I look back at my backlog of things to work on, all of these ideas were also equally exciting.\nWhy is it that new ideas have more weight than old ideas?\nThey say that effectiveness is working quickly on tasks. Efficiency is knowing what to work on. Discipline is the ability to overcome the excitement of a new idea and place it in proper context of all ideas. Sanity is actively saying no to many ideas so you don’t get inundated.\nPerhaps a good strategy is to have a “gut check” for ideas. If they don’t pass the highest threshold then let them go.\nProcess:\n\nWhat’s the new idea?\nWhat’s the potential impact? What would I do with the output of this idea?\nHow feasible is it?\n\nStep 3 is tricky because I don’t want a good idea to go to waste just because it’s hard. Judgment is weighing the trade off between impact and feasibility.\nI like Google Forms for logging ideas. It’s easy to save a Google Form URL to the home screen of my iPhone to make logging new ideas easy. Grooming those ideas is important. Letting go of old ideas is essential. If only I could do the essential…\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "objectID": "posts/2017-10-19-used-cars-analysis/used-cars.html#interpretation-matters",
+ "href": "posts/2017-10-19-used-cars-analysis/used-cars.html#interpretation-matters",
+ "title": "Buying a used car the data science way: Part 2",
+ "section": "Interpretation matters",
+ "text": "Interpretation matters\nFirst, we see the (Intercept). This is the average value of a car for every level of every feature not show. So it represents the average price of a year 2000 with 0 miles, Limited trim, without all wheel drive. This allows us to now see what the average effect of, say, year is, holding all else constant. Every unit is in terms of dollars, so we can see that for as.factor(year)2013 that 2013 cars are worth $1,267 more than 2012 cars. This makes sense.\nI scaled mileage to be in the thousands so that it’s a little easier to interpret. So we see that for every 1,000 miles you drive a Ford Edge, it decreases the value of the car by $91 dollars. This tells you a few things. One, if I’m being offered two cars that are 10K miles different, there should be about a ~$900 difference in the car.\nSecond, if I’m planning on driving my car 30k miles over the next two years, how much should I expect the price to drop just due to mileage? Well, about -$91 * 30 = -$2,730. And maybe if you bought a 2014 model, two years later it could perhaps be like owning a 2012 model today. What’s the two-year difference between a 2014 and a 2012 model? About $3,160 dollars."
},
{
- "objectID": "posts/2022-11-18-writing-a-tech-resume/index.html",
- "href": "posts/2022-11-18-writing-a-tech-resume/index.html",
- "title": "Writing a Tech Resume",
- "section": "",
- "text": "TL;DR: Use the “Accomplished [X] as measured by [Y] by doing [Z]” format.\n\n\n\n\n\n\nWriting resume is painstaking, but let me help jumpstart you. My resume below has gotten me dozens of interviews with top tech companies (Apple, Google, Facebook, Amazon, Microsoft, Waymo, LinkedIn, Hopper, etc).\nIt got me in the door, but let me be clear I didn’t pass most of those interviews 😝. That’s the point of the resume: open the door.\nLet me help you get in the door.\n\nMy template\nHere’s the google doc link to my resume template, also posted below.\nhttps://docs.google.com/document/d/1n8W79UBiWYhMWHeF4sNDtwaWxySRwU3t9isodbUuycI/view\nYou can copy this template and put in your own content.\nI designed and wrote my resume the way I did by following the guidance of these resources: \n\nBlog: Google’s Accomplished X as measured by Y by doing Z formula\nYouTube Video: Google’s How to write a resume\nBlog: Cracking the Coding Interview: how to write a resume\n\n\n\nCommentary\n\nIt takes some practice, but the XYZ formula is very powerful.\nEmboldening words can be helpful to people who have 3 minutes to scan your resume.\nI’ve never gotten commentary on how long my resume is or crammed. I’m guessing people just glance for a few keywords and some decent content and assume the rest of the resume is good.\nI’ve had a lot of discussions in phone calls about my Personal Projects sections. Fellow techies all have their own personal projects and they want to know about mine. It sends a signal you’re intrinsically motivated.\nWhile my resume is long-ish, I try to think: what story does this bullet point tell? What story am I trying to convey?\n\n\n\nHow can I Help?\nI hope this helps! Send me a message at bryanwhitingcoaching@gmail.com if I can help review a resume or give more tips.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "objectID": "posts/2017-10-19-used-cars-analysis/used-cars.html#inference",
+ "href": "posts/2017-10-19-used-cars-analysis/used-cars.html#inference",
+ "title": "Buying a used car the data science way: Part 2",
+ "section": "Inference",
+ "text": "Inference\nNow it’s getting interesting! You’re now starting to remember how your mom told you to never buy a car new, because once you drive it off the lot, it’s worth significantly less. Well, we can see that if you own a brand-new 2017 model and a 2016 model, they’re about $9082 - $6166 = $2916 different. So yea, just by getting the 2017 model you’ll lose $3,000 in the first year. See how 2016 is only $502 dollars more than the 2015? Having the 2015 model is practically the same price as a 2016. So you might as well buy the 2016 model and save a year of wear.\nWell, my theory kind of starts breaking down because 2015 cars and 2014 cars are similarly spaced as 2014 and 2013 ones. So maybe I shouldn’t read too much into this. But the ability to see these trends shouldn’t stop you from asking important questions, like, why is there a difference between years?\nWhen investigating the Nissan Murano, I observed a $5430 difference between 2015 and 2014. That’s non-trivial considering the other years were equally spaced. Doing a little research I saw that Nissan changed the 2015 Murano style, and the new style was apparently worth a lot more. Turns out the same thing happened here for the Ford Edge - the 2015 model is slightly longer with redesigned interior.\nYou can also use this framework to get a decent sense for what features are important to you. Like, is having all wheel drive really worth $1,277? If you live in an area where it snows, it probably is. But if you live in Florida, you may not need it.\nNot only is this framework helpful for comparing within-make-model differences, it helps you get a sense of between-make-model differences. Say, comparing the Ford Edge with the Nissan Murano and the Toyota RAV4. The coefficient for milage on the Edge is -$91. On the Murano it’s -$83, and on the RAV4 -$61. Remember this is the cost per 1,000 miles. That tells me that if I ‘spend’ 30k miles on a car over two years, I’ll lose -$2,730 on the Edge, but only -$1,830 on the RAV4, saving me $900. This also validates that Toyotas hold their value more."
},
{
- "objectID": "posts/2022-11-18-substance-over-style/index.html",
- "href": "posts/2022-11-18-substance-over-style/index.html",
- "title": "Resume Substance over Style",
- "section": "",
- "text": "Earlier in my career I got feedback from a prospective recruiter telling me my resume was “visually imbalanced”. Of course, the recruiter wanted my business.\nBut I went on to get over 70 interviews with that same template. I got a job at FAANG and a top startup using that resume template. I haven’t changed it.\n\nTakeaway\nHow you write your resume and what’s on it will take you farther than how you design it.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "objectID": "posts/2017-10-19-used-cars-analysis/used-cars.html#identify-a-good-market-deal-and-how-truecar-might-do-it",
+ "href": "posts/2017-10-19-used-cars-analysis/used-cars.html#identify-a-good-market-deal-and-how-truecar-might-do-it",
+ "title": "Buying a used car the data science way: Part 2",
+ "section": "Identify a good market deal (and how TrueCar might do it)",
+ "text": "Identify a good market deal (and how TrueCar might do it)\nBeyond the research above, you can use such a model to identify deals in the market and guide your buying decision. Here’s how:\n\nMake a PriceFinder: Get the residuals (actual minus predicted value) for each car. The more negative the residual, the more this car is a good price! This may be how TrueCar and CarGurus get their ‘deals’. They fit a model (probably one better than what I’ve created here) and use its residuals to value each car.\nIdentify ‘overpriced’ cars and try to possibly haggle those dealers down. Again, use the residuals here.\nUse this model to predict ‘out-of-sample’ cars. You can use this model built on TrueCar data and find a car on the classifieds or another site and evaluate whether it’s a good deal or not.\n\nHere’s an example of how we do that. First, let’s predict a hypothetical 2015 SEL Ford Edge FWD with 27.6k miles:\n\nthat_car <- as.data.frame(list(\n year = 2015,\n mileage = 27.6,\n trim = 'SEL',\n awd = 0\n))\n\npredict(mod, newdata = that_car)\n\n## 1 \n## 23520.13\nWe’d expect, on average, such a car to cost $23,520.13. Is what the dealer’s offering you above or below that? If they’re offering $25,000 then you know the car is $1,479.87 higher than what you might expect."
},
{
- "objectID": "posts/2017-02-10-ggplot-plotly/ggplot-plotly.html",
- "href": "posts/2017-02-10-ggplot-plotly/ggplot-plotly.html",
- "title": "The DataViz battle: Plotly vs ggplot2",
+ "objectID": "posts/2017-10-19-used-cars-analysis/used-cars.html#how-to-improve-our-model",
+ "href": "posts/2017-10-19-used-cars-analysis/used-cars.html#how-to-improve-our-model",
+ "title": "Buying a used car the data science way: Part 2",
+ "section": "How to improve our model",
+ "text": "How to improve our model\nNow, this is a pretty naive model, but it’s a model. As George Box said, “All models are wrong, but some are useful.” But here are some quick ways to improve this model:\n\nCheck for interactions. An interaction between miles and year would tell you whether the effect of miles would change across different years. For example, I think adding 10k miles to a 2017 would devalue the car more steeply than adding 10k miles to a 2012 model. During my quick research I checked for interactions, but I didn’t find many to be significant and I chose to ignore them for simplicity.\nUse a linear trend for year rather than a ‘factor’ trend. I’ve chosen here to treat year as a factor, which implies that I think that each year is completely different. Had I used year as a numeric feature, I could have possibly interpreted year as, “Each additional year increases/decreases the price by X on average, holding all else constant.” I liked the simplicity of thinking that each year was different, and since cars usually update models every four years, I wanted to be able to see where this happened. You couldn’t easily see this if year were a linear trend. Objectively, what would be a good way to choose? Fit both models and see which had better R-squared or AIC.\nDo model diagnostics. Check the residual plot to make sure the errors are normally distributed. Check for outliers. Each of these affects the accuracy of your residuals. I did some brief checks and was genearlly satisfied."
+ },
+ {
+ "objectID": "posts/2017-10-19-used-cars-analysis/used-cars.html#takeaways",
+ "href": "posts/2017-10-19-used-cars-analysis/used-cars.html#takeaways",
+ "title": "Buying a used car the data science way: Part 2",
+ "section": "Takeaways",
+ "text": "Takeaways\nSo, a data scientist walks into a dealership with a computer with a registry of thousands of used cars on it. He negotiates with the front-office salesman. And the salesman says…\n\nNon-statistical considerations\nWhen you’re buying a car from a dealership, remember that it’s their job to sell you that car. Here are other random things I learned to think about during the car-buying process.\n\nHow many miles has it been driven, and how old is it? (We found a car with 45,000 miles in one year. That’s almost two trips around the entire world in a year! This is also 3750 miles per month (a cross-country US trip), or about 125 miles per day. What kind of driving do you have to do in a year to get that?\nHow many prior owners were there?\nWas the car previously a rental? A lease? A fleet?\nRental: Abused by X number of people? But also perhaps better maintained by rental companies?\nLease: What kind of person would lease?\nFleet: Probably a corporate car. Like a rental, but different kind of drivers.\nWhere did the car come from?\nIs there snow where they live? And do they use salt on the roads there? (Salt = rust underneath)\nWhat else does the geography tell you about the possible wear on the car (causing long-term problems)\nWhat does the CarFax or Autocheck say?\nWas the car sold at auction? When? What could the time between sales tell you about the car?\nIf someone gets into an accident and doesn’t report it, the CarFax won’t tell you. I drove 45 minutes to look at a ‘great’ deal only to learn the door was nearly falling off.\nGood luck haggling the dealer down. Car dealerships don’t usually make that much on the car, but make their money on warranties and fees. Your best chance is to haggle on the deals being offered (terms and agreements).\nDon’t get the powertrain warranty. This is like the “french fries” of the auto industry. Pure profit.\nEvery dealership has quota. Going in at the end of the month might help you, but the dealership will be packed.\nUsed cars are bought at auction. Carmax buys all their cars at auction and only keeps 1/3 of them. Local dealers also get their cars this way, and they don’t always know what the car looks like till it gets to their lot. Check the Carfax to see if any prior dealers ‘flipped’ the car quickly.\n\nUsed cars have a shelf life. If a car isn’t sold in 45-60 days, some dealerships might have to auction off the car. That means you’ll see a tiering in the price. Less-favorable cars will see the price drop several times in its life cycle. Check CarGurus to see how many times the price has dropped, and try to guess for yourself if it might drop again (or just get sold). They know their prices. They’ve got their back-office guys looking at every price of every same car within 100 mile radius. While I’m also able to get as many of those cars as available on TrueCar, I’m not able to segment by all the various features. It’s easy to webscrape for Make, Model, Year, AWD/FWD, and maybe V6 engine, but it’s very hard to get the details like leather, sunroof, twin-turbo, etc. That’s where the pricing gets really interesting.\n\n\nFinal thoughts\nUltimately, a car is bought by irrational humans, and sold by more experienced ones too. So all of this data will probably go out the door when you try to make a deal.\nSo a data scientist walked into several dealerships, and left frustrated because nobody budged on their prices. Doesn’t matter how much data I have, or what I think I know, everyone’s got their price."
+ },
+ {
+ "objectID": "posts/2023-12-21-github-copilot-levels-up/index.html",
+ "href": "posts/2023-12-21-github-copilot-levels-up/index.html",
+ "title": "GitHub copilot levels up",
"section": "",
- "text": "R users fall in love with ggplot2, the growing standard for data visualization in R. The ability to quickly vizualize trends, and customize just about anything you’d want, make it a powerful tool. Yet this week, I made a discovery that may reduce how much I used ggplot2. Enter plot_ly().\nFor this post, I assume that you have a working knowledge of the dplyr (or magrittr) and ggplot2 packages. I caveat that this post is backed with only 4-5 hours using plotly(), so some statements here may not be fully vetted.\nPlotly and ggplot2 are inherently for different purposes. plotly allows you to quickly create beautiful, reactive D3 plots that are particularly powerful in websites and dashboards. You can hover your mouse over the plots and see the data values, zoom in and out of specific regions, and capture stills. Here’s a basic histogram:\n\nset.seed(1)\nlibrary(plotly)\nplot_ly(x = ~ rnorm(1000), type = \"histogram\")\n\n\n\n\n\nAfter a brief dabble this week in plotly, I realized quickly the many advantages that plotly has over ggplot2.\nSeveral initial impressions:\n\nPlotly handles multiple wide data columns. I always find it annoying that to color different series in ggplot2, your data had to be in long format. Granted, it takes one simple melt() command to get the data into wide format.\nPlotly also handles long format (see below).\nCustomizing the layout (plot borders, y axis) is easier.\nCustomizing the legend is easier (in ggplot2 I’ve wanted to remove just one series, which isn’t always easy).\nDocumentation is better in Plotly.\nPlotly syntax is very intuitive (learning how aes() in ggplot2 works is tricky at first)\nPlotly also works for Python, Matlab, and Excel, among other languages.\nIt’s very easy to add new series and customize them (one line, one scatter, and one bar, for example)\nYou can use other fonts (which is possible in ggplot2, but I’ve never gotten to work on my Windows machine)\nYou can toggle series on and off by clicking the series name in the legend\n\nBenefits of ggplot2 over plotly:\n\nFacet wrapping is very easy in ggplot2. (I think you have to do subplots in plotly.)\nggplot2 is probably quicker for exploratory analysis.\n\n\nConverting ggplot2 into plotly\nAn additional benefit of plotly is that you can convert your ggplot() graphs into a plotly object.\n\nlibrary(ggplot2)\np <- qplot(x = rnorm(1000), geom = \"histogram\")\np\n\n\n\n\n\n\n\n\nThen, invoking the ggplotly(p) command, we see the transformation:\n\nggplotly(p)\n\n\n\n\n\nA draw back of ggplotly() is that if you do refined customization (like putting your legend on the bottom of the graph), ggplotly() doesn’t seem to pick this up by default.\n\n# ggplot with legend on the bottom\np <- qplot(\n data = iris,\n x = Sepal.Width,\n y = Sepal.Length,\n geom = \"point\",\n color = Species\n) +\n theme(legend.position = \"bottom\")\np\n\n\n\n\n\n\n\n# Plotly doesn't pick up the legend change\nggplotly(p)\n\n\n\n\n\nBut since Plotly also saves to an object, you can use the %>% notation to pipe and add additional plotting commands. This is similar to the + operator in ggplot().\n\np <- qplot(\n data = iris,\n x = Sepal.Width,\n y = Sepal.Length,\n geom = \"point\",\n color = Species\n) +\n theme(legend.position = \"bottom\")\np2 <- ggplotly(p)\n# Use the plotly layout() command for legend customization\np2 %>% layout(legend = list(orientation = \"h\"))\n\n\n\n\n\nThe legend doesn’t do exactly what we want, but you can manipulate the legend location manually using x and y coordinates. The orientation = 'h' setting in the docs puts the legend on the bottom for default plot_ly() objects. Graphing the same series, we see the legend at the bottom:\n\nplot_ly(iris,\n x = ~Sepal.Width,\n y = ~Sepal.Length,\n type = \"scatter\",\n mode = \"markers\",\n color = ~Species\n) %>%\n layout(legend = list(orientation = \"h\"))\n\n\n\n\n\n(You notice the Plotly X-axis title can get cut off1, so let’s put that +1 to ggplot2.)\n1 This may depend on your screen.Plotly seems very intuitive relative to ggplot2 in doing layout customization. Things that took me many iterations on StackOverflow to figure out, like adding a black line on y = 0, are built in to Plotly.\n\np <- plot_ly(iris,\n x = ~Sepal.Width,\n y = ~Sepal.Length,\n type = \"scatter\",\n mode = \"markers\",\n color = ~Species\n)\n# Put legend on bottom, change the x-axis range, and turn on the x-axis line. \n# Also, make the zeroline visible, and turn it red.\np <- p %>% layout(\n legend = list(orientation = \"h\"),\n xaxis = list(\n zeroline = T, # Turns x = 0 on\n zerolinecolor = \"red\", # colors x = 0 red\n showline = T, # Shows xaxis border line\n range = c(-2, 7)\n )\n)\n# Or, save parameters into a list. Use new fonts (a huge plus)\nf1 <- list(\n family = \"Arial, sans-serif\",\n size = 18,\n color = \"lightgrey\"\n)\nyax <- list(\n title = \"Sepal length\",\n titlefont = f1\n)\np %>% layout(yaxis = yax)\n\n\n\n\n\nThings I’d like to further explore:\n\nYou can export static plotly images out to file. My hypothesis is that Plotly images take longer to generate than ggplot2. So if I’m mass producing 30,000 plots (which I had to do last month), which is the faster approach? I would assume ggplot2.\n\n\n\nPlotly in RShiny Dashboards\nThe goal in learning Plotly was for me was to eliminate the Excel-VBA dashboard I created using for my manager. Excel has (some) benefits over ggplot2 static charts: you can easily hover your mouse over a series to see the data value, and most industry users know how to manage an Excel axes. Grated, you can build in an RShiny widget to allow the user to control the axes, but Excel comes with that knowledge base built-in. ggvis allows for the powerful library of Google charts, but I think for a reactive dashboard, plotly is a great way to go2.\n2 Of course, this is relative to the chart you’re trying to make.So Plotly solved the Excel problem for me. Now my manager can click and zoom to the parts of the graph that are interesting, and hover the mouse to see the values. Just use renderPlotly() instead of renderPlot() in the server.R file, and plotlyOutput() instead of plotOutput() in the ui.R file.\nMore info here: RShiny and Plotly\n\n\nRShiny vs Plotly Dashboards\nBoth RShiny and Plotly allow for creating dashboards. Plotly allows you to build dashboards as well. If you’re just interested in only visualizing charts and trends, Plotly dashboards seem like the way to go. But to build reactivity into your dashboard (like subsetting your sample, changing date ranges, etc.), RShiny still seems like the more customizable solution.\n\n\nFinal thoughts\nOverall, it seems that ggplot2 is quicker to build and find what you want. With facet wrapping, the qplot() command, and ggsave(), you can whip something up fast. Plotly is better for dashboards, as you can interact with the plots. I feel like Plotly has a better syntax and documentation, and so it may be easier to get a basic plot to look how you want it to. But ggplot2 seems to have more advanced features, so if you want to get into refined customization, you may want to stick with ggplot2. They’re both great, and serve different purposes, but I’ll be using plotly for my RMarkdown and RShiny visualizations going forward.\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "TL;DR:\n\n\n\n\n\n\n\n\ncan debug failed terminal errors\nshows git diff of revisions\ncan explore workspaces now with @workspace (still lags behind Cursor, which indexes everything in your code base)\nauto-commit messages (I’ll take this)\nauto-pr messages (sounds good to me)\n\nMore here:\nhttps://code.visualstudio.com/blogs/2023/11/13/vscode-copilot-smarter\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2022-12-15-chatgpt-rshiny/index.html",
- "href": "posts/2022-12-15-chatgpt-rshiny/index.html",
- "title": "Building a complex R Shiny Dashboard Using ChatGPT",
+ "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html",
+ "href": "posts/2022-06-23-the-best-todolist-manager/index.html",
+ "title": "What is the best todo list manager?",
"section": "",
- "text": "ChatGPT is nothing short of amazing, but I think I’ll still have a job as a data scientist for at least another year or so.\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Wirecutter reviewed the best todo list apps and declared “ticktick” the winner. After all my research here’s what I’ve learned.\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2022-12-15-chatgpt-rshiny/index.html#data-processing",
- "href": "posts/2022-12-15-chatgpt-rshiny/index.html#data-processing",
- "title": "Building a complex R Shiny Dashboard Using ChatGPT",
- "section": "Data Processing",
- "text": "Data Processing\nI asked it to create an indicator variable for me so that I could color my dygraphs chart by forecasts vs. historical:\n\nHere was my prompt:\n\nDon’t use add_series() since you’re wrong and it’s not a function in dygraphs. Instead, append the forecasts to the aapl in the stock_subset function and create a new indicator for the forecasted values where 0 is past data and 1 is forecasted data. Then, color the dygraphs chart based on the indicator value. Re-write this script below accordingly.\n\nLook at that code! It knew to filter appl2$ds > max(appl$ds). That’s exactly how you subset a prophet dataframe (which returns all historical and future forecasts in one dataframe). It knew that appl was the first dataset and appl2 was the derived dataset. It knew that it had to take the max date from appl. This is absolutely amazing, hands down.\naapl2 <- bind_rows(aapl2, forecast_subset)\naapl2$indicator <- ifelse(aapl2$ds > max(aapl$ds), 1, 0)"
+ "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#structured",
+ "href": "posts/2022-06-23-the-best-todolist-manager/index.html#structured",
+ "title": "What is the best todo list manager?",
+ "section": "Structured",
+ "text": "Structured\nThe highest rated app for planning on the App Store is “Structured”. Where this app excels is it’s simplicity. It’s likely best for someone like a student that wants to plan out when they’ll get their homework done. What I loved about this app, and what I wish many others apps had, is a duration estimation for a task. Why is it so hard for developers to realize that tasks take time? They obsess over list keeping and app design but ignore the second most basic question: how long will this take? Structured kind of solves this. What I don’t like about this app is the design. While others praise it for its beauty and simplicity I thought the visual feel of it was distracting."
},
{
- "objectID": "posts/2022-12-15-chatgpt-rshiny/index.html#data-inspection",
- "href": "posts/2022-12-15-chatgpt-rshiny/index.html#data-inspection",
- "title": "Building a complex R Shiny Dashboard Using ChatGPT",
- "section": "Data inspection",
- "text": "Data inspection\nI then got this error:\nI'm getting this error: Error in `select()`:\n! Can't subset columns that don't exist.\n✖ Column `date` doesn't exist.\nChatGPT gave me the ring-around. It didn’t know that the dates were stored as the rownames, but confidently gave me answers like this:\n\nTo it’s credit, it recommended I inspect the data myself:\n\nOnce I figured out the problem, I told it that the rownames were dates, and that date wasn’t a column. It had just the solution for me:"
+ "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#microsoft-to-do",
+ "href": "posts/2022-06-23-the-best-todolist-manager/index.html#microsoft-to-do",
+ "title": "What is the best todo list manager?",
+ "section": "Microsoft To Do",
+ "text": "Microsoft To Do\nMicrosoft To Do used to be a favorite of mine. I loved its daily planner and smart suggestions. It has the philosophy that every day you should start fresh. All unfinished tasks will be removed from today and you’re left to pick and choose what you want to do. The fatal limitation however is that it lacks the ability to plan tomorrows a tasks the day before. So this forces people to have to populate their tasks at the start of the day. This is the worst time for planning for me because I want to get up and get going ASAP. (But honestly most times I wake up 15 minutes before my first meeting or am spending time pouring milk into my kids’ cereal.) Morning is not the best time to plan a day for me, so I lost interest here.\n\n\n\nTo Do has smart text parsing.\n\n\n\n\n\nTo Do has “smart suggestions” that you can add to your day."
},
{
- "objectID": "posts/2022-12-15-chatgpt-rshiny/index.html#misleading-functions",
- "href": "posts/2022-12-15-chatgpt-rshiny/index.html#misleading-functions",
- "title": "Building a complex R Shiny Dashboard Using ChatGPT",
- "section": "Misleading Functions",
- "text": "Misleading Functions\nTwice, it told me that dygraphs had functions that it didn’t actually have. It even wrote code for me. For example, it recommended this to me:\ndygraph(stock_subset(), main = \"AAPL Stock Price\") %>%\n add_series(data = forecast_subset, col = \"red\") %>%\n dyRangeSelector()\nAnd it doubled down:"
+ "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#todoist",
+ "href": "posts/2022-06-23-the-best-todolist-manager/index.html#todoist",
+ "title": "What is the best todo list manager?",
+ "section": "Todoist",
+ "text": "Todoist\nI’ve tried many times to get into Todoist. It has smart text parsing which is apparently best in class from my Reddit research.\nThings it has:\n\nProjects can have section labels. This lets you organize your list more meaningfully. This is great, only that TickTick does it better. Each project in TickTick can have subprojects. These subprojects act like sections. You can also move subproject from one list to another.\n\n\nThings it lacks:\n\nCan’t nest labels (tags)\nCan’t create labels dynamically - you need to manually add them on the label screen before a task can leverage them.\nCan’t pull in my calendar.\nDoesn’t have check items, only subtasks."
},
{
- "objectID": "posts/2022-12-15-chatgpt-rshiny/index.html#data-conversion",
- "href": "posts/2022-12-15-chatgpt-rshiny/index.html#data-conversion",
- "title": "Building a complex R Shiny Dashboard Using ChatGPT",
- "section": "Data Conversion",
- "text": "Data Conversion\nIt told me to use the xts package so that dygraphs could read the data in. This was accurate - dygraphs cannot read in tibbles.\n\nBut it was misleading me a little bit. Again, I had to look at the data, because the direct code it gave me was missing columns that it assumed it had."
+ "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#sorted3",
+ "href": "posts/2022-06-23-the-best-todolist-manager/index.html#sorted3",
+ "title": "What is the best todo list manager?",
+ "section": "Sorted3",
+ "text": "Sorted3\nThis app takes a different design. It’s main selling point is the ability to auto schedule your day based on the tasks you have for that day.\nPros:\n\nRich formatting on description. Supports markdown.\nclean UI. Good shortcuts.\nintegrates with calendar to plan your day.\nEasily plan duration.\n\nReally nice widgets.\n\nCons:\n\nDoesn’t have subtasks, but has check boxes\nNo prioritization or smart filters. Just tags and lists.\nCan’t easily see across lists like you can in TickTick\n\nAuto schedule is nice but could be kind of gimmicky. If you have 8 meetings in a day and five 30 min free blocks but all your tasks for the day are an hour, it would put all your tasks at the end of the day. I found it kind of buggy."
},
{
- "objectID": "posts/2022-11-12-what-is-a-model/index.html",
- "href": "posts/2022-11-12-what-is-a-model/index.html",
- "title": "What is a model?",
- "section": "",
- "text": "_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#automated",
+ "href": "posts/2022-06-23-the-best-todolist-manager/index.html#automated",
+ "title": "What is the best todo list manager?",
+ "section": "Automated",
+ "text": "Automated\nMotion became popular as a calendar organizer. Add a task and Motion will figure out when to do that for you. If you change your meeting then Motion will adjust when your tasks will get done. Cons: it’s expensive ($20-$35 per month). It’s mobile app is horrid. It didn’t seem like it offered sub tasks and all that. So you really need to trust the algorithm.\n\nIt gets the notion of task-duration right. But if you want to jot down notes of all the next actions you need to take on a project, your up the creek. I’m also not sure how it handles task dependency - if I put in 5 tasks for a project will it know their order? Not sure. Again I’ve never tried this app so take this all with a grain of salt.\nReclaim is another app claiming to do the same thing and offers a free tier. I did as much as set up my account with this.\nPerhaps if I gave these more attention they’d prove their worth. I’ve tried calendaring my todos in the past and it hasn’t been very effective for me. Perhaps these apps would solve it but I think I need a good mobile app.\nAnyway, you can’t beat free and free is working for me on TickTick right now. Also, my calendar is very open generally so I don’t have the same problems a busy manager might have.\nI think these apps would be best for people who have meeting heavy workflows. It handles the schedules as you need."
},
{
- "objectID": "posts/2022-11-12-what-is-a-model/index.html#im-building-a-model-you-know",
- "href": "posts/2022-11-12-what-is-a-model/index.html#im-building-a-model-you-know",
- "title": "What is a model?",
- "section": "I’m building a model, you know",
- "text": "I’m building a model, you know\nThe first time I heard the word “model” was from an engineering professor in college. I had no idea what he meant so I asked him. He looked at me like many who have forgotten their own ignorance and just said “Some calculations in excel. You know.”\nNo, you don’t know. But nod your head so you don’t look like an idiot. “Right, of course.”\nWhy this moment was intriguing to me is one of my personal mysteries. But it was one of those odd moments that stuck. I’m now here to help my old self make sense of the world."
+ "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#akiflow",
+ "href": "posts/2022-06-23-the-best-todolist-manager/index.html#akiflow",
+ "title": "What is the best todo list manager?",
+ "section": "Akiflow",
+ "text": "Akiflow\nThis app seemed expensive ($15/mo), no mobile app, but has an awesome calendar. The philosophy is you have to assign every task to a date and time to move it out of the inbox. That seemed like a headache to me, and seemed like I’d be constantly managing the inbox. That and I’m cheap and want a mobile app."
},
{
- "objectID": "posts/2022-11-12-what-is-a-model/index.html#the-same-name-for-different-things",
- "href": "posts/2022-11-12-what-is-a-model/index.html#the-same-name-for-different-things",
- "title": "What is a model?",
- "section": "The Same Name for Different Things",
- "text": "The Same Name for Different Things\nWhen I told my mom I build models for a living she shook her head. Turns out my dad had been telling her for years that he builds financial models, yet she never figured out exactly what it was that he did.\nLet’s be clear on the common ground. A “model” per the dictionary is “simplified description, especially a mathematical one, of a system or process, to assist calculations and predictions.” By this definition, my dad and I do the same thing. We use numbers to put structure on the world."
+ "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#sunsama",
+ "href": "posts/2022-06-23-the-best-todolist-manager/index.html#sunsama",
+ "title": "What is the best todo list manager?",
+ "section": "Sunsama",
+ "text": "Sunsama\n$20 per mo or $192/year. Super fast customer support.\nFeatures: * Mobile app, Mac app. * Duration\nPros: * Channels (lists) * great shortcuts * can align tasks with objectives * more of a mindful daily planner * Integrates with Todoist and other apps. You can pull in all to foist metadata including subtasks and check them off in Sunsama. Additionally, you can add duration and calendar your Todoist tasks. * Can export all the tasks to csv. * Can split calendar events.\nCons: * Widgets are weak * No priority labels * No task tags/labels. (But you can use Todoist for this if you want.)"
},
{
- "objectID": "posts/2022-11-12-what-is-a-model/index.html#types-of-models",
- "href": "posts/2022-11-12-what-is-a-model/index.html#types-of-models",
- "title": "What is a model?",
- "section": "Types of Models",
- "text": "Types of Models\nA financial model is different than a statistical model is different than a physical model or a 3D architecture model. But they all do the same thing: use numbers to put structure on the world.\nEvery model has a few things in common: - They make assumptions about the world - They use data - They probably make some calculations to help someone make a decision\nA financial model might make assumptions in a spreadsheet about how many customers you’ll get per month for the next 12 months. This doesn’t seem complicated, and it shouldn’t: models don’t have to be complicated.\nA data science model is one that might try to predict whether a customer is going to click on an ad."
+ "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#reclaim.ai",
+ "href": "posts/2022-06-23-the-best-todolist-manager/index.html#reclaim.ai",
+ "title": "What is the best todo list manager?",
+ "section": "Reclaim.ai",
+ "text": "Reclaim.ai\nTask manager isn’t as good as Sunsama. Only $8/mo for pro. Has todoist integration too, but that’s a pro feature.\nThis is one I’ll want to try later. It has the habit tracking and smart flexible calendaring. Not sure how good of a task manager it is. TBD."
},
{
- "objectID": "posts/2022-11-12-what-is-a-model/index.html#how-does-a-model-work",
- "href": "posts/2022-11-12-what-is-a-model/index.html#how-does-a-model-work",
- "title": "What is a model?",
- "section": "How does a model work?",
- "text": "How does a model work?\nThe purpose of a math model is to imitate the “data-generating process”. Meaning, if you have some data on your customers (what they purchased and when), you would hope that there’s some rules that your customers are operating under. If you knew their income, their budget, their preferences, etc then you can guess how much they have to spend at a given time. Thus, if you know how much gas is in their tank, the mileage of the vehicle, the MPG of the vehicle, you can predict when they’ll stop to fill up on a road tip. When they fill up, they generate some data. “Bob purchased $34.52 of gas on Monday at 5:43pm.” Why did they purchase at that time of day? And why did they spend exactly $34.52?\nIf all you have outcome/results data then you can’t ever make predictions. But if you know what type of car Bob drives, how often he drives, how much gas he had in the morning when he left his house, etc, you can start to predict when he’ll next fill up.\nThe assumption behind a model is that what was true in the past will remain true in the future.\nWhy have there been so many issues because of COVID? Nobody could have predicted a world wide pandemic. The world wasn’t the same. All the models stopped working. Supply chain models. Financial models. Default models. All of these assumptions across all industries started breaking down."
+ "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#vimcal",
+ "href": "posts/2022-06-23-the-best-todolist-manager/index.html#vimcal",
+ "title": "What is the best todo list manager?",
+ "section": "Vimcal",
+ "text": "Vimcal\nFast calendar. Not for tasks or backlogs it seems though. Didn’t try."
},
{
- "objectID": "posts/2022-11-12-what-is-a-model/index.html#how-do-you-build-a-data-science-model",
- "href": "posts/2022-11-12-what-is-a-model/index.html#how-do-you-build-a-data-science-model",
- "title": "What is a model?",
- "section": "How do you build a data science model?",
- "text": "How do you build a data science model?\nMost people build models every time they calculate the “mean”. This is the most ubiquitous model and anybody can use it. Sum your revenue and divide by number of purchases.\nIf you want to predict revenue for next Monday, just take the average revenue for every past Monday. That’s a model.\nIf Monday is a holiday or the Monday after Black Friday, your model might be very wrong. So you estimate black Friday’s sales by looking at last year’s sales. “Last Cyber Monday we sold 2x more than we typically do on a Monday. So I’ll predict that on this upcoming Cyber Monday I’ll do 2x more than what I did in the most recent Monday”\nNow that you’re taking in some data, your building a model. You’re translating a theory on how the world works into numbers. You’re taking last year’s sales and using it to predict tomorrow’s sales."
+ "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#asana",
+ "href": "posts/2022-06-23-the-best-todolist-manager/index.html#asana",
+ "title": "What is the best todo list manager?",
+ "section": "Asana",
+ "text": "Asana\nOh man, how many times have I downloaded Asana? I started with it in 2014 and thought it was cool. Trying it again in 2022 it had a similar feel.\nAsana is the only task manager app I reviewed that has dependency management. Sure, Jira has blockers and all that but Jiras more aimed toward high level project management. It’s not fast at the\nOther things I remember liking:\n\ninfinite subtasks. This was fun and also distracting.\nAutomation features. You can create workflows for what happens to a task\nCommunication. It has a nice inbox feature that let me communicate with my manager easily.\n\nCons:\n\nClunky on the board management. If I want to start a new project things get messy and overwhelming."
},
{
- "objectID": "posts/2022-11-12-what-is-a-model/index.html#start-simple",
- "href": "posts/2022-11-12-what-is-a-model/index.html#start-simple",
- "title": "What is a model?",
- "section": "Start Simple",
- "text": "Start Simple\nWhenever I build models, I always try to start simple and see if something more complicated can beat it.\n\nUse averages\nBuild a rules-based model\nBuild a regression model/machine learning model\nBuild a ML model with rules on top of it\n\nIf I’m trying to predict whether a user will click an ad, the simplest model is “what % of all users clicked this ad in the past?” If 20% of all users who see an ad click on it, then my “model” is “predict a 20% chance of click”.\nThen to make it more complicated, you can segment your users. “10% of men click on this but 25% of women do.” You now have more data and better predictions.\nA “decision tree” is a machine learning algorithm that just finds those segments in your data. “If man age 30 in Chicago then 22.3% chance of clicking based on all past behavior of all 30-year old men in Chicago”. It’s called a decision tree because it will put every user into mutually exclusive groups.\nSome “business logic” model might be making predictions around how your business operates. For example, if there’s a sale you may know that clicks increase."
+ "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#jiragithub-projects",
+ "href": "posts/2022-06-23-the-best-todolist-manager/index.html#jiragithub-projects",
+ "title": "What is the best todo list manager?",
+ "section": "Jira/GitHub projects",
+ "text": "Jira/GitHub projects\nGreat at a high level but messy to manage the gritty day to day. Doesn’t let you schedule or put due dates. So eventually I’d need some other tool to manage all the minutia of getting things done."
},
{
- "objectID": "posts/2023-12-15-prompt-techniques/index.html",
- "href": "posts/2023-12-15-prompt-techniques/index.html",
- "title": "Prompt techniques",
- "section": "",
- "text": "How to Prompt Engineer\n\n\n\n\n\n\nTL;DR: Prompt Engineering\n\n\n\n\n\n\n\n\nOpen AI released a great Prompt Engineering guide (link in comments). Here is a summary of their 6 strategies for getting better results when prompting GPT-4 pic.twitter.com/QiBASsbOHD\n\n— MindBranches (@MindBranches) December 15, 2023\n\n\nThe summary above is from here:\nPrompt Engineering\n\n\nTakeaways\nIf prompt engineering is the future of coding, then coding is going to change from a deterministic, imperative experience to a stochastic one. That seems odd. 95% chance your output code will do what you want it to do, with a 5% change it does something drastically wrong. How do you unit test that?\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "objectID": "posts/2022-06-23-the-best-todolist-manager/index.html#not-considered",
+ "href": "posts/2022-06-23-the-best-todolist-manager/index.html#not-considered",
+ "title": "What is the best todo list manager?",
+ "section": "Not considered",
+ "text": "Not considered\nI didn’t look at Things3 or OmniFocus, which are considered of the best tools. Things3 was too expensive and seemed like it had many of the same features as TickTick, except I personally didn’t like the UI."
},
{
- "objectID": "posts/2018-07-02-semantic-versioning-for-data-science-models/semantic-versioning-for-data-science-models.html",
- "href": "posts/2018-07-02-semantic-versioning-for-data-science-models/semantic-versioning-for-data-science-models.html",
- "title": "Semantic Versioning for Data Science Models",
+ "objectID": "posts/2023-12-13-learn-by-doing/index.html",
+ "href": "posts/2023-12-13-learn-by-doing/index.html",
+ "title": "Learn by doing",
"section": "",
- "text": "If you’ve ever wanted to tag your data science model, you’ve probably wondered how to version it. Which will it be: vx.4.1, v34.1231.51.21, or v91.x4.dev34? After reading about semantic versioning, I propose a method for versioning data science models.\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "TL;DR: Yes, unintelligent people can use things without fully understanding them\n\n\n\n\n\n\n\n\nMany people have their noses up their ass when they claim you can't use an algorithm you don't fully understand.They give themselves this fake moral superiority. They must be the only ones who belong because they know every detail.This is all bullshit.Here are a few simple…\n\n— Santiago (@svpino) December 12, 2023\n\n\nWhile the language is perhaps a bit strong, I agree with the sentiment here.\nI don’t have a PhD but have been limited twice in my career by those who wish I had one. What would having a PhD give you that I don’t have?\nWell, I trust you more if you had one.\nJeremy Howard, a world renowned research and creator of fast.ai and answers.ai started his lectures on deep learning with “you don’t need to have a math degree, you just need to know a little bit of code”. With that philosophy he’s helped entrepreneurs create amazing products that left his free online program.\n\nBlindspots tho\nConcepts are important to understand. Understanding your training data and how the machines learn from them and the biases they’re prone to is very useful, perhaps mission critical, to creating value.\nSo what’s the line? Well, if you got a PhD you’d probably spend 80% of your time on your dissertation and taking some non relevant coursework. How does that help you today?\nWell, you should learn as many concepts as you can. Like driving a car, you don’t need to know hydroplaning but you should learn hydroplaning. You don’t need to know how to drive in snow but you should know that driving a car in the snow with rear wheel drive is a different experience.\nIf you live in Arizona you don’t need to know how to drive in the snow. Your bias is you’ll never know why that would be necessary.\n\n\nPrinciple\nBe so curious that you explore the boundaries of your space such that you learn “oh, cars work differently in snow, but I don’t have snow so I’m safe”. Once you reach the boundary you probably know enough. Then ship your code and learn by how you messed things up.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2018-07-02-semantic-versioning-for-data-science-models/semantic-versioning-for-data-science-models.html#an-example",
- "href": "posts/2018-07-02-semantic-versioning-for-data-science-models/semantic-versioning-for-data-science-models.html#an-example",
- "title": "Semantic Versioning for Data Science Models",
- "section": "An Example",
- "text": "An Example\nI build data science models by building python packages and committing the code to GitHub. The python package contains all of the support files and a main.py file to run the pipeline from start to finish. The support files could contain either helpful loggers to tell me where the pipeline is breaking down, diagnostic tools such as an AUC-ROC plot, files to engineer features, or files to train different model types. So there are really these three things that could happen to my code at any time. It doesn’t matter where those changes occur, or how many lines of code changed. What matters is what’s happened to the model. Each time I get a pull request approved, I’ll update the version number in my repo.\n\nChanging BETWEEN\nIncrement the BETWEEN version for the following changes:\n\nWhen the structure of your data changes\nIf your target variable changes (how you coded it, or data that produced the target changed)\nThe underlying population you’re training on changes\n\nAssume I discovered that I had a bad join. The join change might have been a minor one, but it’s now difficult to really compare the two models because my target was affected as a result. What I was predicting, even if slightly different, is now something different. If you ever feel like you’re comparing apples to oranges when looking at an AUC-ROC curve, update the BETWEEN version.\nWith each BETWEEN change, there should be a clear communication in the release notes of why the model is inherently different from prior models. These changes should be less frequent.\n\n\nChanging WITHIN\nIncrement the WITHIN number when the following happens:\n\nFeatures are added\nData sources are added or updated. (You might have several data sources today that help contribute to the target. But adding a data source might just mean adding new features. If you add data in such a way that it changes what you’re ultimately modeling, update the BETWEEN version.)\nNew modeling types are added. (You might be using a logistic regression, but add a mo)\n\nSo if I added files that allowed me to train different model types, or added files that engineered new features, I’d incrememnt the WITHIN version. In my mind, this shouldn’t be a count of features you include, but should just be incremented every time you do something that affects the model performance.\nAt any time, you might what to view different subsets of features and their effect on the model. So, you might not change the number of features, but add some functionality into your pipeline that now produces three different models in each run. I would increment the WITHIN version number in that case.\nRemember, the goal of this type of versioning is to show generally what’s changed in your pipeline.\n\n\nChanging PROCESS\nIncrement the PROCESS version when the following happens:\n\nA minor bug is fixed (unless this updates BETWEEN or WITHIN)\nEnhance a logger or aspect of the pipeline\nAdd a diagnostic plot or table\n\nBasically, any time you add something that doesn’t ultimately affect the performance of the model, you should update the PROCESS version. Your changes might improve the flow of your code and make you a much happier developer, but if it doesn’t improve the model, then the key priority of your code hasn’t improved."
+ "objectID": "posts/2023-12-25-ai-robots-are-mastering-tasks-faster-than-humans-/index.html",
+ "href": "posts/2023-12-25-ai-robots-are-mastering-tasks-faster-than-humans-/index.html",
+ "title": "AI Robots are mastering tasks faster than humans",
+ "section": "",
+ "text": "With a proper objective function, machines can do really hard, isolated tasks\n\n\n\n\nthey can walk\nthey can solve complex physical challenges\nthey can play the toughest strategic games on earth and win (AlphaGo 2017, AlphaStar 2019)\nbut can they determine the objective function by themselves?\n\n\n\n\nReinforcement Learning\nGreat article written at Techradar going over the advances in reinforcement learning that allow a robot to finish labyrinth faster than humans. The video does the feat even more justice:\n\nThis stands out to me in a few ways, principally that they crippled the machine to not take shortcuts. It still won.\nReminds me first of AlphaGo:\n\nWhich reminds me of AlphaStar. Here’s a video with commentary on AlphaStar’s performance:\n\nThe thing I connect this video (or some other I watched) with labyrinth is that AlphaStar devised brand new strategies. Players against AlphaStar remarked how they felt like it was indeed being creative. Likewise, Cyberrunner (labyrinth) found shortcuts.\nAll of this points to reinforcement learning as a viable solution to getting stuff done both in the digital and physical worlds. Take, for example, how RL is used to teach a robot how to walk.\n\n\n\nAn AI robot learning to play StarCraft for the first time…\n\n\n\n\nWhat’s the objective?\nThe goal for Super AI will be to figure out what the objective is. If we can marry digital and physical worlds, teaching the systems how to solve objective functions, the possibilities become…super/human.\nThe unique thing about a human, it seems, is that we have our objective function hardwired into us: survive. First this is “food/shelter”. Next it’s social survival.\nHow do humans figure out objective functions so fast? A human looking at labyrinth will know what to do within a second. Does it have interest/patience to sit there and learn it? Does GPT 4Vision have the ability to figure out these objective functions?\nWe learn by objective functions. Minimize wrong answers on a test. Minimize missed baskets on the basketball court. Minimize rejections in a sales funnel. Maximize great conversation and connection at a dinner party.\nHumans learn very quickly. For now, robots are being isolated to specific tasks and their results are impressive. But perhaps to become our overlords all they need is the one objective function to rule them all. Er, uhm, and by “them” I of course don’t mean “us”…\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2020-12-18-productivity-notes/index.html",
- "href": "posts/2020-12-18-productivity-notes/index.html",
- "title": "Productivity notes",
+ "objectID": "posts/2022-07-03-when-data-collection-goes-too-far/index.html",
+ "href": "posts/2022-07-03-when-data-collection-goes-too-far/index.html",
+ "title": "When data collection goes too far",
"section": "",
- "text": "Time Management\n\nCalendar my ToDos.\nShorter work is better. Less to edit. Less mistakes. Keep it simple.\n\n\n\nDistractions\n\nBlock 30 min for email and review team members work twice a day.\nAggressively filter email spam.\n\n\n\nAnalysis\n\nStart with a story. End with story.\nIterate quickly with stakeholders.\nalways have one nights sleep before sending a presentation. Edit.\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Self Surveillance\nI came across this article in Flowing Data and thought it was fascinating.\nI’ve been on a productivity kick lately and notice these apps have habit trackers. I’ve tried using them in the past and have realized it gets too much too quickly. But that doesn’t mean as a data guy that I don’t think it’s a cool idea. It just overwhelms me.\nWe’ll I discovered that me tracking a few habits a day is junior varsity level. Felix’s blog is how it’s done to the extreme. Clearly this guy isn’t worried about Big Tech or Big Government harvesting his personal data. I love the boldness!\nThis is just 1/20th of the full page screen shot:\n\nThe number of slices and views he has is endless. It’s crazy to see everything he’s tracking, partly I’m sure because he has fun with it and the other part because he deep down hopes there’s something there.\nHis quote on Flowing Data has stuck with me:\n\nOverall, having spent a significant amount of time building this project, scaling it up to the size it’s at now, as well as analysing the data, the main conclusion is that it is not worth building your own solution, and investing this much time. When I first started building this project 3 years ago, I expected to learn way more surprising and interesting facts. There were some, and it’s super interesting to look through those graphs, however retrospectively, it did not justify the hundreds of hours I invested in this project.\n\nThis guy probably knows himself in a way that none of us ever will. He can’t ever lie to himself and say “yea, I’m good at exercising every day” because he’s got the data to back it up. But is this the life we want?\n\n\nSelf Surveillance as a Service\nSSaaS? Apparently. Exist.io tries to find patterns in your personal data and tell you when your most happy.\n\n\n\nExist.io\n\n\nWhen I researched the Apple Watch team a few years back I came to appreciate what they were/are trying to do: correlate heart monitoring with insurance claims data to proactively identify preemptive signs of cardiac arrest and other health issues.\nThat seems meaningful in theory because information about other people can be correlated with information about me.\nBut does info about myself correlate with info about myself? Probably. I have cycles. I have common traps and pitfalls like any human walking this earth. Some pitfalls are daily, some are quarterly or seasonal.\nBut I can also take 5 minutes to ponder reflectively or talk to a close friend and identify those patterns. What’s more human?\nAnd what’s more: if you end up changing your behavior then your past can’t really predict your future. So all of the data becomes moot once it achieves its purpose: true lasting change. Because if you truly change your behavior based on the insights you glean about yourself, those past triggers and anomalies no longer affect you in the way they did. So when you experience a personal step change in life, you’re a new you. Maybe not. But maybe.\nAlso, the new generations growing up on smartphones have a unique challenge: their online identities are given to them before they’re really conscious enough to determine if that’s what they want to show the world. Parents posting YouTube or TikToks of their kids are cementing for these kids an identity they themselves didn’t choose. We’ve all gone through phases of deleting Facebook photos from 15 years ago because it doesn’t reflect our current self image. What’s to be said of giving people an opportunity to change if everything they’ve ever said or done is recorded somewhere? Twitter. Instagram. TikTok. Facebook. Every post is cementing us in a way.\nI think it’s better to forget yesterday. Because it doesn’t have to have any bearing on where you’re going tomorrow. It can if you let it. But your past doesn’t have to define you.\n\n\nTakeaways\nWould you rather spend hours setting up digital automatic and manual tracking systems just to learn what you could by opening a gratitude journal every day? Sometimes less is more. I don’t think digital self surveillance is a direction we should go. (Not to mention the moment you do get hacked you’ll deeply regret it.)\nEven blogging can be dangerous for your career. Never know what’s gonna come back to haunt you years later.\nSo I should be careful what I say here, that is if I ever care to work for a self surveillance company in the future.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2023-12-12-future-of-coding/index.html",
- "href": "posts/2023-12-12-future-of-coding/index.html",
- "title": "Future of Coding",
+ "objectID": "posts/2023-12-19-how-to-encourage-your-kids-/index.html",
+ "href": "posts/2023-12-19-how-to-encourage-your-kids-/index.html",
+ "title": "How to encourage your kids",
"section": "",
- "text": "The Essence\n\n\n\n\n\n\nTL;DR: The future belongs to those whom it’s always belonged\n\n\n\n\n\n\nA blogger data scientist Santiago recently said on Twitter, (paraphrasing) “I’m going to focus on what won’t change”. He was worried about how coding and AI will be automated, but Jeff Bezos wasn’t worried about so much of the world changing with the Internet. Bezos focused on what wasn’t changing.\nThis thought hit me, because I think it’s relevant to what our future looks like in 18 months.\n\n\nEvery day, a new winner\nNothings more apparent that the world is going to be wildly different than the past than to hop on X and see new academics one upping themselves with the next greatest LLM. Mistral just announced a 2Bn valuation as a 6 month old company. Fast is the new pace. And it’s oddly the case that big businesses are capable of massive disruption from upstarts. They can’t afford to stay glacial much longer (5-10 years).\nWith all of this, what about this blog post won’t change? Well, the timeless principle is that the nature of the game has changed permanently. Pandora’s box was opened.\n\n\nThe rise of a new equalizer\nGo into a poor part of town and you’re guaranteed to see one thing: poor people using smart phones.\nThey have access to all the same information that the rich people have. The same excellent software. Same excellent hardware. All for a fee of $50/mo for the hardware and $25/mo for an unlimited data plan.\nSo that out the rich and the poor on the same level.\nWell now, knowledge and reasoning is as accessible as Google Maps. You don’t have to pay $20/mo for ChatGPT plus. You can use Bing, Claude.ai, bard.google.com, You.com, or perplexity.ai for free. Now.\nSure people hate on bard and want to use the latest and greatest. But even the crappiest of the four I mentioned above (granted Bing/You.com use GPT in the background), is still leaps ahead of where we were just a year ago. If we were stuck only using one of these tools forever we’d still be massively more productive than not.\nSo knowledge and intellectual reasoning is going to be commonplace amongst the plebes.\n\n\nif you value intelligence above all other human qualities, you’re gonna have a bad time\n\n— Ilya Sutskever (@ilyasut) October 7, 2023\n\n\n\n\nMore of the same\nSo…what’s going to happen? Well, I’m not sure human nature will change much.\nThey say “being rich doesn’t make you happier. It just makes you more of what you already were before being rich”.\nAnd so, the mentally poor will remain so. They’ll use AI for entertainment. Better video games. Better, faster content. More consumption.\nThe creatives will be more creative. Nobody will be at a barrier of being silo’d. So what if you’ve only done data science for 10 years. You know coding. It’s easy to now do web design. Or copy writing. Or art. Or…\nAnd I won’t be able to stop the masses from using Code Interpreter to do data science. The systems will only become more powerful, so cloud companies will make it easier to write sql or whatever. Heck, 5-10 years English will be the only programming language. (Who here does firmware/bit mapping anymore? Only a select few.)\n\n\nThe hottest new programming language is English\n\n— Andrej Karpathy (@karpathy) January 24, 2023\n\n\nPython programmers will become like the engineers who create the seeds. And the rest of us just get bigger, better, and more efficient farms. There will always be those who specialize, as there are people who write compilers today, but fewer and farther between. Perhaps intellectual interest and economic incentives will equalize.\n\n\nFestivus for the rest of us\nCompanies will become smaller. But the same human needs will need to be met:\n\nfood\nshelter\nclothes\nphysical toys\ndigital entertainment\n\nMore people will just be able to do more things. Perhaps there will be more prosperity as a result.\nAnd more people will unionize and try to prevent the change. But that’s a short term thing. Can’t last forever. A union is only as good as it’s needed.\n\n\nTakeaways\nGet ahead by focusing on what’s essential. Leverage what’s around you to develop new skills to prepare for the future. And be excited. It’s going to be a wild ride.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Foster talent by:\n\ncreating stress free environment\nfinish before they’re tired\ndon’t pamper them at the country club\ntreat them for 10 minutes after\nno need to teach. Kids will figure it out\npeople who are great don’t necessarily love what they do, but those who love what they do will be great\n\n\n\nThis from Padraig Harrington should be required viewing for parents. Some really great advice.pic.twitter.com/ExE5rZpg7a\n\n— Kendall Baker (@kendallbaker) December 18, 2023\n\n\nI love it.\nThis is probably good management advice in any capacity.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/quarto-from-phone/index.html",
- "href": "posts/quarto-from-phone/index.html",
- "title": "How to blog from your phone using quarto (or hugo/jekyll/etc) using Working Copy and Obsidian",
+ "objectID": "posts/2020-12-16-pretotyping-ml-solutions/index.html",
+ "href": "posts/2020-12-16-pretotyping-ml-solutions/index.html",
+ "title": "How to mock the value of an ML solution",
"section": "",
- "text": "Here are the steps I use to post content from my phone to bryanwhiting.com.\n\nCreate a new, empty vault in Obsidian\nOpen working copy, set up synced repo. This means any changes are auto-synced.\nAdd remote, point it at your GitHub repo where your blog is\nSet up a “pull, stage, commit, push” shortcut\nSet up GitHub actions to build, and push (use my repo’s workflow.yaml if you need an example)\nHave GitHub pages (if your repo is private) or Netlify (if your repo is private, because you can’t have a private GitHub repo without paying)\n\nAnd that’s it! Write markdown in obsidian. Push when you’re ready, let GitHub actions build and deploy. Check GitHub actions logs for any errors! (You’ll get an email if it fails.)\n\n\n\nSet up synced repo\n\n\n\n\n\nLink the repo\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "IBM once launched a demo of speech-to-text in the early day’s of the technology. But there was a twist: it was smoke and mirrors. A typist in behind the stage was translating and typing.\nThis is a mechanical Turk.\nIn data science, we can present results or dashboards to audience members before they see anything. If you see a big reaction, you know you found a number they cared about.\nGo ahead, write the conclusion first. Fake the chart first. Then back out the code that could produce the insights they need.\nSometimes, unique insights appear along the way. But most of the time you can properly gut-check whether someone will care about what you’re doing. Save time. Work backward.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2023-12-21-the-dodo/index.html",
- "href": "posts/2023-12-21-the-dodo/index.html",
- "title": "The Allegory of the Dodo",
+ "objectID": "posts/2022-06-21-products-for-dads/index.html",
+ "href": "posts/2022-06-21-products-for-dads/index.html",
+ "title": "Products for dads",
"section": "",
- "text": "The scoop\nThere’s a great kids book, called “There’s a dodo on the wedding cake”.\nSo there’s this dodo, right, who walks in and hops on a cake. The attending violinist then scares it away and is praised for protecting the cake.\nTo the reader’s delight, more animals start appearing and the violinist excitedly protects the cake through hilarious misadventure. (My favorite is when he says, “begone beavers!”)\nTwo funny things happen though:\n\nThe guy soon realizes that all these creatures are wedding guests.\nThe guy has such a big nose and he’s allergic to marshmallows, that when he’s offered some cake he sneezes and the cake goes flying.\n\n\n\n\ndodo on a cake\n\n\n\n\nMorale?\n\nWhen you praise someone for doing something good, they do it again and again. → There’s a natural hero inside all of us.\nBut only the first animal was a problem. The beavers were the chefs. So he ends up creating all these problems thinking they’re problems. → In tech, when you fix a bug you then think other things are bugs and go off fixing those too. No, just do the first. Then check in.\nThe irony is that, despite saving the cake, he sneezes and ruins the cake. → Despite all the effort, the greatest protection to the cake was him staying away from it. Sometimes by trying to solve a problem we can make things way worse and create all these problems. (Politics? Create problems by creating solutions to problems that need solutions.)\n\nDon’t mistake a beaver for a dodo. But don’t mistake a dodo for a wedding guest.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Products I enjoy as a dad\nThird times a charm - I hope by(/if) kid number four I’ll have this all figured out. It’s amazing how I feel so new to it all despite raising two kids in the last five years. In starting fresh a third time I feel inspired to write down some things I’m rediscovering.\nTo anyone getting started on their baby journey, perhaps these suggestions can help. Your baby is definitely different than mine, so your mileage may vary! But since I’m still using this stuff five years later maybe it can help you too. Either way, congrats and good luck!\n\nProducts I still use on baby 3\nSince I’m a mega influencer (on my very established data science blog here), I get paid $500 every time you click a link below. So please, go crazy.\nI’ll also credit my wife for figuring out to buy these products in the first place.\n\nStretchy swaddle, like this. I like stretchy swaddles because they’re forgiving. The cotton muslin swaddles are nice and soft, but if the fabric doesn’t stretch the right way you may have to redo the swaddle.\nOllie swaddle (nighttime). This thing is ridiculously overpriced for a swaddle. $70? Don’t get it. But if you do, I like it because its super easy to put on and very sturdy (hard for baby to break). At month 2 the baby gets strong and the basic swaddles are so easy to break. That and I feel like I’m gonna hurt my kid if I try to make a cotton swaddle unbreakable. Enter Velcro! I can be gentle on the swaddle but have it be unbreakable. There are a lot of random contraption swaddles that tie the baby up like a pretzel. This is your classic burrito. Insert baby, wrap right side, Velcro left side and you’re done.\nUbbi Steel Odor Locking diaperpail. I also hated the thought five years ago of shelling out $70 for a trash can. Five years since and the thing still performs. I know how well it’s working when garbage day rolls around and I open it to change it. Smells like a gentle coastal breeze, maybe. To this day my wife hasn’t changed it more than a time or two because thankfully I have an iron stomach and unfortunately she does not. Point is: it keeps the diaper smell inside (for the most part) and uses normal garbage bags (cost saver).\nBaby K’tan Original Baby Wrap Carrier, Infant and Child Sling. I’ve tried several carriers. Boba wraps take way too much patience to learn how to put on (patience I ran out of quickly as a new dad). This thing is just two circular fabric swaths, which means it takes two seconds to throw on. The convenience is huge - baby crying? Grab the carrier and snuggle up in 10 seconds or less.\nDockATot Deluxe+ Dock. We just got one off Facebook marketplace for $50 for third kid. We got it because let’s be real - sleeping is hard. The hardest thing about sleeping is the uncertainty what’s happening on the other side of the room in the crib. Grunting. Coughing. Is baby waking up? Needs a pacifier? When it’s 5am and you’re on feeding three of the night, being able to just roll over and plop the little one into the co-sleeper and monitor the situation close up is nice.\n4moms rockaRoo Baby Swing. There are a thousand ways to solve the “where do I put my baby while cooking dinner” problem. We’ve liked this one.\n\n\n\nMust Haves\n\nDish soap. Honestly, nothing gets out a blowout stain like dish soap. That and drying in the sun is a miracle worker. (Apparently this is how cloth diaper people do it.)\nApple cider vinegar for diaper rash. Once the little pimples appear, you’ll try anything to get them to go away. We just discovered this hack recently and I was delightfully surprised how easily it worked.\nThis is the right way to carry a car seat. A friend of mine that has three kids didn’t know this was a thing until he saw me do it. It’s a thing. Do it! Your forearm will thank you.\n\n\n\nToddler Time\nDown the road, these products have hit home runs for me.\n\nMunchkin Miracle 360 Trainer Cup. Every time I put this cup together I consider applying to Munchkin for a job to be their sales rep. This company has very clever designs on their products and this one just crushes it for me. Acts like a normal cup. Performs like a sippy. (I’ll admit I’ve taken a few swigs out of this just to understand how it works!)\nContigo Spill-Proof Kids Tumbler. I think we have 80 water bottles in a junk drawer somewhere. Cheap ones. Expensive ones. They all end up scattered in pieces like the zebra herd before the lion that is your innocent child. This cup is no exception: I’ve lost a straw on one already. But! If you want a water bottle you’ll have the problem of spare parts. (Uhh, this isn’t a recommendation so far…) So why this cup? This cup is just good. I don’t know. It’s easy to assemble and clean (wide top). Doesn’t spill. Kids like straws?\n\n\n\nBooks\nNot that you’ll have time to read…\n\nThe Happiest Baby on the Block; Fully Revised and Updated Second Edition: The New Way to Calm Crying and Help Your Newborn Baby Sleep Longer. I’m sure there’s a quick YouTube video showing the 5 ways to sooth a baby. (Here’s an article.) But if you like books this is a good one. TLDR: sucking, swaddle, shushing, swinging, and side are all soothing to a crying baby.\n1-2-3 Magic: Gentle 3-Step Child & Toddler Discipline for Calm, Effective, and Happy Parenting. This is for toddlers, but I’ve learned this technique has been very helpful for giving me sanity in the terrible two phase. The approach doesn’t always work, but I’ve learned when it does, which is important. TLDR: don’t talk (engage in debate) with your toddler or show emotion when they’re misbehaving. Count them when they do bad acts. That’s 1. That’s 2. That’s 3. Break time. The book explains the method. YouTube is probably a friend here too.\n\n\n\nFinal word\nFatherhood has pushed my limits in all the ways. But I thank God every day I’m a dad. I do a nightly gratitude journal of my favorite moments of the day. If I’m paying attention during the day, at night it’s always my kids that have brought me the happiest moments and greatest joys of the day.\nJust remember: the clothes/products/trips are like the tools used to construct a fine painting. At the end of the day, it doesn’t matter what tools you use. Nobody cares and neither will you. It’s the output that matters most. I’ve found by just really paying attention to and getting curious about my kids in any moment helps me see the beauty of it all, while distractions cause me to think I need this or that be a better dad.\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/writing-is-enlivening/index.html",
- "href": "posts/writing-is-enlivening/index.html",
- "title": "Writing is truly enlivening",
+ "objectID": "posts/quarto-feature-requests/index.html",
+ "href": "posts/quarto-feature-requests/index.html",
+ "title": "Things I’d like in a blog",
"section": "",
- "text": "Hamilton is running out of time\n\n\nI’ve been writing for my whole life. I have at least 15-20 journals. I have 3000 notes in one app, 2200 notes in another, at least 1800 digital journal entries over 10 years, I’ve built and replaced probably 5 blogs. This blog has 57 entries so far. What have I learned?\n\nMake writing daily as easy as possible. I set up some automation so I can write from my phone for daily blogging.\nWriting makes things feel real for me. It helps connect the dots. It’s a far more productive way to read the news. Less stuff, more thinking. Because if I can’t write at least one useful note about something I read, I realize it wasn’t worth reading.\nPerhaps reading is the more important daily duty. Well, to write I need something to inspire me. Usually that comes from the news. Sometimes inspiration comes from yesterday’s post. This inspiration came from editing my prior post.\n\nI’m not gonna worry about who reads this. But instead focus on writing every day because…\n\nI think it helps me practice storytelling\nhelps me practice communication\n\nHamilton wrote like he was “running out of time”. I’m writing because I want to use my time how I enjoy it.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "quarto has strengths and drawbacks\n\n\n\nOther frameworks (like Hugo) are more flexible. But quarto offers a lot more out of the box and handles the stuff I really don’t want to have to think about.\n\n\n\n\n\nA traditional Quarto\n\n\nThings quarto is good at:\n\nHandles metadata really well (easy to get a title image, description, Twitter card, etc.)\nheavily customizable: can organize posts as a list, as a table, as a grid, etc. can easily customize what’s in that. The main thing I didn’t like about Hugo was that the style was just slightly different than I liked it.\ncomments\nsearching\ncategories/tags\neasy to create different/custom listings\ncan blog from my phone using Working Copy and GitHub actions\n\nThings wish quarto were better at:\n\nFaster render (Hugo speeds plz)\nDon’t render future posts\nA blog landing page that had a stream of entries, rather than just a list. (Similar to seths.blog)\nListings based on tags\nmake it easy to have an alternate slug (make all posts at root, and not in posts/)\n\nThings I’d like to do with my blog:\n\ngenerate weekly summaries\nbuild some subscribers\nsend out a weekly summary of content.\nauto-post to Twitter, Instagram (using photo + description + zapier) and Facebook.\nauto-post anything with “LinkedIn” tag to LinkedIn\nmove editing into a vault in Obsidian\nmake it easier to copy ChatGPT photos to my blog post folder\n\nFor now, what it offers widely covers what it doesn’t. Most blog templates don’t allow for search, custom listings, lists of tags, Twitter content, etc. For now, quarto offers 95% of what I need.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2022-12-23-who-do-you-want-to-be-known-for/index.html",
- "href": "posts/2022-12-23-who-do-you-want-to-be-known-for/index.html",
- "title": "How do you want to be remembered?",
+ "objectID": "posts/2022-06-22-downstream-impact/index.html",
+ "href": "posts/2022-06-22-downstream-impact/index.html",
+ "title": "Downstream Impact",
"section": "",
- "text": "On the sports fields of high school, I was often asked, “How do you want to be remembered?” The energy behind that question is a rallying cry for someone to go out and score some goals and be victorious. Score goals → be remembered forever in the annals of high school lore, the thinking goes.\nThis question misses the point.\nIn “Ego is the Enemy”, Ryan Holiday asks the rhetorical question “to be or to do”. The mindset is that either we can focus on doing great work (spending time on action and education), or we can focus on being great (talking too much, dreaming too much). Ironically, greatness follows those who do, not those who say they will do.\nSo instead of asking “how will others remember me”, ask “what will I do for others?”\n\n\n\nDan Kiefer via Unsplash\n\n\nOn this Christmas season, remember that while Jesus Christ’s divinity has been debated for centuries, certainly he’s been remembered. He’s been remembered because of the actions he took and the kindness he showed.\nChristians believe he was the Son of God, one with literal power over death and rose from his own grave. He could calm the seas and turn water to wine. He was the rightful heir to the throne that his Roman captors sat on. He could have easily freed his oppressed nation to rest on his laurels forever.\nYet the recorded miracles we have from him don’t involve him showing moving mountains or collecting power and gold. The reason he’s remembered is because His miracles involve him spending time with individuals and in their service. He healed the sick. Caused the blind to see. He alleviated suffering.\nMay this Christmas be one where you don’t focus on what others think of you. May it be one where you enjoy following His example of doing good for others, however small.\n\nWhosoever will be great among you, let him be your minister; And whosoever will be chief among you, let him be your servant: Even as the Son of man came not to be ministered unto, but to minister, and to give his life a ransom for many. Matthew 20:26–28\n\nMerry Christmas.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "Every pea is unique. But it’s still a pea.\n\n\nOnline experiments are powerful tools to see the causal impact of a business strategy. Tweak something, see what users do.\nThe limitations of online experiments is that once you manipulate your user experience there’s no going back. Users are forever impacted by your choice.\nIf you give out a credit card with $100, $300, and $500 lines, people are going to spend more on the $500 line. They might also default more because they’re less able to pay back $500 than they are $100. As a result, you may lose these $500 users forever and might impact other aspects of their life, not to mention impact the marketing department’s strategy.\nA group of users is like a single living organism, if you treat it like that. If you only measure one metric (conversion rate) on the entire massive audience, then you’ll optimize for the “average” user.\nBut nobody is average. We are all distinct. We fall into categories, sure. But to really optimize the whole business we need to optimize for pockets of users. Or, we just need to set up a system that’s personalized.\nTakeaway? Personalize your experiments where possible. Consider the ethical and downstream impacts of the experiment.\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/write-despite-ai/index.html",
- "href": "posts/write-despite-ai/index.html",
- "title": "Write despite AI",
+ "objectID": "posts/2017-10-18-used-cars-scrape/used-cars-scrape.html",
+ "href": "posts/2017-10-18-used-cars-scrape/used-cars-scrape.html",
+ "title": "Buying a used car the data science way: Part 1",
"section": "",
- "text": "Write because you can\nI don’t write because anyone is actually listening. But if I’m lucky, the AI is actually listening.\n\nIf AI is listening (webscraping me), I’m fortunate to teach it. Essentially, good training data is hard to come by, especially as AI content proliferates on the internet. Let’s consider it a privilege I’m able to help it learn.\nWriting and telling stories are the oldest form of humanity. Even if an AI can do it better, perhaps we’ll just spend time doing more of it.\n\n\n\n\nWhen plume meets tablet\n\n\nSo yea, I’ll write despite AI. I’ll write because it’s very human to do so.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "[Update 2021-11-16] This analysis was originally written on my old blog here. You can find the source code for it here. This code no longer works as TrueCar changed their CSS to make it more difficult to scrape. It’s still possible, but you’d need to build a custom scraper from scratch.\nThis is part 1 out of a two-part series on scraping used car data. Check out part 2 to learn how to analyze the data.\nIn another post, I describe how I use this data that I’ve scraped, but I wanted to provide a more in-depth tutorial for those interested in how I got the data. Note, this data belongs to Truecar, so all uses herein are for personal and academic reasons only.\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
},
{
- "objectID": "posts/2022-06-28-increase-focus-through-writing-down-everything/index.html",
- "href": "posts/2022-06-28-increase-focus-through-writing-down-everything/index.html",
- "title": "Increase Focus through Writing Down Everything",
- "section": "",
- "text": "The GTD framework preaches something to the effect of “get every todo item down and into a system. It may take dozens or hundreds of tasks to get it all out.” Essentially, if a task is in your mind, it will slow you down. But if you put it down on paper, it will free your mind to think on other things. Also, more ideas will come.\nI’ve put perhaps a hundred things down on paper in the last few days. Interestingly, ideas I had forgotten about came back to my mind. I wrote these down.\nThere got to a point today though where I had so much on my screen I felt overwhelmed. How will I ever sort through all this mess?\nThankfully, most of the things don’t need to get done and are either “eventually” or “someday maybe” tasks, and not “asap” tasks. So just ignore them.\nAnd secondly, I always “know” what’s most important. This documentation process doesn’t change that. It just helps me to get all the distractions out of my head and down on paper.\nSpending time organizing all this stuff seems like a potential waste. But what I realize is that it improves my just-in-time recall. If earlier in the day I planned to buy a dish scrubber in the evening, when the evening rolls around that task magically comes back to my mind. So it’s really just front loading all the decision making and prioritizing. I think our brains are capable of more subconscious processing than we realize.\n(Pause entry to buy dish scrubber…) okay, I’m back.\nAnother idea that’s having an effect on me is the 2 min rule. If a task takes less than 2 min, don’t organize it. Just do it. Powerfully, this mindset helps me to care less about two min tasks. Or, sometimes there’s a tendency to over exaggerate the stress something will cause. Like buying a scrubber could be a 10 min task if you want to find the perfect one. But if you only allocate yourself 2 min, you realize this is only a $6 decision and I have better things to do with my time.\nLastly, I love being deliberate about tasks by writing down the intended outcome of a task. It’s very easy as a data scientist to get swallowed up in all the weeds. Or, rather, when your task is to dig up a gem it’s easy to get distracted by all the weeds that need to be pulled that are in your way. So by being clear about the reason I’m diving into the data in the first place helps me re-surface less scathed. Data are a dirty business…\n\nTakeaways\nThe effect of the GTD framework?\n\nLess FOMO. By writing down everything I can see clearly that 1) I don’t have time for all of it and 2) the stuff I care about pops up.\nI don’t have to worry about what I’m not doing because I can easily say no to it as I say yes to something better.\nI can see broader connections between a litany of things. This helps me prioritize what my subconscious has been telling me by nagging me about X ideas.\nIf I start working on something not in my list it makes me ask if this thing is fealty worth my time when I know I have a ton of other stuff to do. (Recency bias I spoke about in my other blog post.)\n\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "objectID": "posts/2017-10-18-used-cars-scrape/used-cars-scrape.html#get-the-data",
+ "href": "posts/2017-10-18-used-cars-scrape/used-cars-scrape.html#get-the-data",
+ "title": "Buying a used car the data science way: Part 1",
+ "section": "Get the data",
+ "text": "Get the data\nIn order to do any good analaysis, you first need data. I prefer to have more data than less, where possible. In this case, I don’t have any data, so I use webscraping to get the data. There are much better tutorials on how to scrape data, so I’ll be light. I use R’s rvest package here, which does a decent job.1 Let’s look at Truecar’s Used Car postings2. First I use google to find the search query on Truecar that I like."
},
{
- "objectID": "posts/2022-06-28-pomodoro-principles/index.html",
- "href": "posts/2022-06-28-pomodoro-principles/index.html",
- "title": "Pomodoro Principles",
- "section": "",
- "text": "I’ve been trying a Pomodoro timer in TickTick and it’s changing how I approach my work day. I’ve been using it for two days so far and here are the differences I notice:\n\nWhen my mind knows I’ll get a break at 25 minutes, it allows me to focus more.\nIt’s easier to say not to slack messages or impulses to open the news if I know I’ll have time for those things later.\nA 5 min break is perfect.\nIf a distraction pops up, I write it down in my journal and tell myself I can follow up during my break.\n25 minutes is short. It’s amazing how time flies.\n\nI used to hate timing myself after working in consulting. I vowed I would never work on the clock ever again. But don’t the best athletes pace themselves? Any marathon trainer knows how to set a pace and have rest periods. Why wouldn’t the same principles apply to cognitive work?\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "objectID": "posts/2017-10-18-used-cars-scrape/used-cars-scrape.html#load-packages",
+ "href": "posts/2017-10-18-used-cars-scrape/used-cars-scrape.html#load-packages",
+ "title": "Buying a used car the data science way: Part 1",
+ "section": "Load packages",
+ "text": "Load packages\n\nlibrary(rvest)\nlibrary(dplyr)\nlibrary(magrittr)\n# Find the URL of the data you want to scrape\nurl <- 'https://www.truecar.com/used-cars-for-sale/listings/ford/edge/'\nread_html(url)\n\n## {xml_document}\n## <html lang=\"en-US\">\n## [1] <head>\\n<meta http-equiv=\"Content-Type\" content=\"text/html; charset= ...\n## [2] <body>\\n <!-- Element target for any partner code meant to execut ...\nYou’ll see there’s a head and a body. Our data’s in the body, so let’s use html_nodes() and html_text() to parse out the data we want. I used Selectorgadget to know what HTML classes to search for.\n\nread_html(url) %>% html_nodes('.col-xs-6.col-sm-8.no-left-padding') %>% html_text()\n\n## character(0)\nSo that’s how you get the data on a single page. If you look closer at the URL, you see a lot of helpful things. First, there’s the make, then the model, then the location-zip, then the year-range, and ultimately the trim. This is a very pretty and clean URL. If you click on a few additional pages, you’ll see the URL opens up with ?page=2.\nhttps://www.truecar.com/used-cars-for-sale/listings/ford/edge/location-90210/?page=2\nThis is our ‘in’ to scraping multiple pages. I won’t bore you with the details of how to get that data into a neat matrix for us to analyze, but suffice it to say that I’m able to do it. Just build a function to construct a URL, and build a loop to go through the different pages, then use lots of str_extract() from the stringr package and gsub to clean up the data.\n\nlibrary(stringr)\n\nmake = 'ford'\nmodel = 'edge'\nzip = '90210'\nyear = 2012\nnpages = 5\n\nurl <- paste('https://www.truecar.com/used-cars-for-sale/listings/', \n make, '/', \n model ,\n '/location-', zip,\n '/year-',year,'-max/?page=', sep = \"\")\n\nurls <- paste(url, 1:npages, sep = \"\")\n\nscrape <- function(pageno){\n try(\n read_html(urls[pageno]) %>% html_nodes('.col-xs-6.col-sm-8.no-left-padding') %>% html_text()\n )\n}\n\nlong_list = scrape(1)\nfor(i in 2:npages){\n print(i)\n new_list = try(scrape(i))\n \n error = (\"try-error\" %in% class(new_list))\n \n if( error == FALSE ){\n long_list = c(long_list, new_list) \n } else {\n break\n }\n}\n## [1] 2\nstats <- long_list\ndf <- as.data.frame(stats)\ndf$stats %<>% as.character()\ndf$price <- str_extract(df$stats, '\\\\$[0-9]*,[0-9]*') %>% \n gsub('Price: |\\\\$|,', '', .) %>%\n as.numeric()\ndf$year <- str_extract(df$stats, '^[0-9]* ') %>% \n as.numeric()\ndf$mileage <- str_extract(df$stats, 'Mileage: [0-9]*,[0-9]*') %>% \n gsub('Mileage: |,', '', .) %>%\n as.numeric()\n\n# a = df$stats[1]\ndf$trim <- str_extract(df$stats, '.*Mileage:') %>% \n gsub('FWD|AWD|4x[24]|[24]WD|V6|4-cyl|^[0-9][0-9][0-9][0-9]|4dr|Automatic|Manual|Mileage:', '', ., ignore.case = T) %>% \n gsub(make, '', ., ignore.case = T) %>% \n gsub(model, '', ., ignore.case = T) %>% \n trimws() \n\n\ndf$awd <- grepl('AWD|4WD|4x4', df$stats, ignore.case = T) %>% as.numeric()\ndf$manual <- grepl('manual', df$stats) %>% as.numeric()\ndf$v6 <- grepl('V6', df$stats) %>% as.numeric()\ndf$location <- str_extract(df$stats, 'Location: .*Exterior:') %>% \n gsub('Location: |Exterior:', '', .) %>% \n trimws() \ndf$ext <- str_extract(df$stats, 'Exterior: .*Interior:') %>% \n gsub('Interior:|Exterior:', '', .) %>% \n trimws() \ndf$int <- str_extract(df$stats, 'Interior: .*VIN:') %>% \n gsub('Interior: |VIN:', '', .) %>% \n trimws() \ndf$vin <- str_extract(df$stats, 'VIN: .*\\\\$') %>% \n gsub('VIN: |\\\\$', '', .) %>% \n substr(., 1, 17)\ndf$deal <- str_extract(df$stats, '\\\\$[0-9]*,[0-9]* below') %>% \n gsub('below|\\\\$|,', '', .) %>% trimws() %>%\n as.numeric()\n\nAnd here’s what the results look like. You’ve got the original scraped data in the stats column and then everything else that you can parse out.\n\n# df was the dataframe object we needed\ndf %>% select(-stats) %>% head(10) %>% formattable::formattable()"
},
{
- "objectID": "posts/2022-06-21-the-power-of-party-pay-by-visible/index.html",
- "href": "posts/2022-06-21-the-power-of-party-pay-by-visible/index.html",
- "title": "The Power of Party Pay by Visible",
+ "objectID": "gratitude.html",
+ "href": "gratitude.html",
+ "title": "Gratitude",
"section": "",
- "text": "Visible’s party pay madness\nVisible wireless offers $25 off to users who join a party. So a user who signs up to pay $50 for unlimited data can click one button to join a party and only have to pay $25/mo from then on. Why not just offer users $25 to begin with?\nIt’s really confusing to me as a user - why not just give me the $25 automatically if it takes nothing to join a group? Because of gamification.\n\n\nWhy it works\nI think their strategy is as follows:\n\nThey hope some people get brought in on marketing and never remember to activate a party pay. Let’s say this number is 20-40%. Perhaps these users fund the rest of the discounts.\nJoining a party pay group for free is an easy way to make your customers win. I.e., a user gets $25 off for joining. Gimmicky, but I think it works. So a user has an immediate sense of gratification - a great way to build a strong emotional bond. “Click a button to get $25 off in perpetuity? Nice!” Anyone would agree this feels better than just starting off with $25.\nThis makes users want to share and tell others. Especially if customers feel they’ve gamed the system. Who today doesn’t like a hack?\n\nBut even if (1) is wrong, and that 95% of people join a party I’m sure they’ve done the math to consider how many people they need to sign up to make the money work.\n\n\nThe Larger Strategy\nWhy would Verizon cannibalize their own sales? I switched my wife out of a $70 unlimited plan on Verizon to a $25 plan on visible for the same service. Well some customers will stay with Verizon because of its prestige and the add on benefits. But users who are cutting the cord are going somewhere else, I’m thinking Verizon wants to stay ahead of the curve. It’s a race to the bottom, sure. But it doesn’t hurt to try to win the race.\n\n\n\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only."
+ "text": "“Gratus sum, ergo habeo.” (I’m thankful, therefore I have.)\nIn November 2020, a Christian apostle named Russell Nelson invited the world to flood social media with gratitude. I rather enjoyed posting gratitude publicly, so here we go.\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nI’m grateful for artists\n\n\nThose who create, who learn, who think in new ways, who pioneer, who explore.\n\n\n\n\n\n\nDec 27, 2023\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nI’m grateful for blankets\n\n\nMy grandma made blankets cool\n\n\n\n\n\n\nDec 26, 2023\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nI’m grateful for doctors\n\n\nDoctors of the body. Doctors of the mind. Doctors of the soul…Any engaged in the pursuit of healing.\n\n\n\n\n\n\nDec 25, 2023\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nLight\n\n\nI set up 72 lights today.\n\n\n\n\n\n\nDec 24, 2023\n\n\n\n\n\n\n\n\nNo matching items\n\n_________________________ Bryan is an amatueur father, Christian, and tech writer, but a professional data scientist. To learn more, see bio. To get more, subscribe. Weekly/monthly emails only.",
+ "crumbs": [
+ "Gratitude"
+ ]
},
{
"objectID": "sortspec.html",
diff --git a/docs/sitemap.xml b/docs/sitemap.xml
index ca9cdd776..13f1aa68a 100644
--- a/docs/sitemap.xml
+++ b/docs/sitemap.xml
@@ -2,266 +2,278 @@
https://www.bryanwhiting.com/christianity.html
- 2023-12-28T06:15:26.333Z
+ 2023-12-28T06:46:34.273Z
- https://www.bryanwhiting.com/gratitude.html
- 2023-12-28T06:15:26.525Z
+ https://www.bryanwhiting.com/little-lamb.png.html
+ 2023-12-28T06:46:34.497Z
- https://www.bryanwhiting.com/posts/2017-10-18-used-cars-scrape/used-cars-scrape.html
- 2023-12-28T06:15:26.565Z
+ https://www.bryanwhiting.com/posts/2022-06-21-the-power-of-party-pay-by-visible/index.html
+ 2023-12-28T06:46:34.537Z
- https://www.bryanwhiting.com/posts/2022-06-22-downstream-impact/index.html
- 2023-12-28T06:15:26.577Z
+ https://www.bryanwhiting.com/posts/2022-06-28-pomodoro-principles/index.html
+ 2023-12-28T06:46:34.561Z
- https://www.bryanwhiting.com/posts/quarto-feature-requests/index.html
- 2023-12-28T06:15:26.681Z
+ https://www.bryanwhiting.com/posts/2022-06-28-increase-focus-through-writing-down-everything/index.html
+ 2023-12-28T06:46:34.561Z
- https://www.bryanwhiting.com/posts/2022-06-21-products-for-dads/index.html
- 2023-12-28T06:15:26.573Z
+ https://www.bryanwhiting.com/posts/write-despite-ai/index.html
+ 2023-12-28T06:46:34.645Z
- https://www.bryanwhiting.com/posts/2020-12-16-pretotyping-ml-solutions/index.html
- 2023-12-28T06:15:26.569Z
+ https://www.bryanwhiting.com/posts/2022-12-23-who-do-you-want-to-be-known-for/index.html
+ 2023-12-28T06:46:34.601Z
- https://www.bryanwhiting.com/posts/2023-12-19-how-to-encourage-your-kids-/index.html
- 2023-12-28T06:15:26.665Z
+ https://www.bryanwhiting.com/posts/writing-is-enlivening/index.html
+ 2023-12-28T06:46:34.645Z
- https://www.bryanwhiting.com/posts/2022-07-03-when-data-collection-goes-too-far/index.html
- 2023-12-28T06:15:26.621Z
+ https://www.bryanwhiting.com/posts/2023-12-21-the-dodo/index.html
+ 2023-12-28T06:46:34.633Z
- https://www.bryanwhiting.com/posts/2023-12-25-ai-robots-are-mastering-tasks-faster-than-humans-/index.html
- 2023-12-28T06:15:26.677Z
+ https://www.bryanwhiting.com/posts/quarto-from-phone/index.html
+ 2023-12-28T06:46:34.641Z
- https://www.bryanwhiting.com/posts/2023-12-13-learn-by-doing/index.html
- 2023-12-28T06:15:26.649Z
+ https://www.bryanwhiting.com/posts/2023-12-12-future-of-coding/index.html
+ 2023-12-28T06:46:34.609Z
- https://www.bryanwhiting.com/posts/2022-06-23-the-best-todolist-manager/index.html
- 2023-12-28T06:15:26.589Z
+ https://www.bryanwhiting.com/posts/2020-12-18-productivity-notes/index.html
+ 2023-12-28T06:46:34.533Z
- https://www.bryanwhiting.com/posts/2023-12-21-github-copilot-levels-up/index.html
- 2023-12-28T06:15:26.665Z
+ https://www.bryanwhiting.com/posts/2018-07-02-semantic-versioning-for-data-science-models/semantic-versioning-for-data-science-models.html
+ 2023-12-28T06:46:34.529Z
- https://www.bryanwhiting.com/posts/2017-10-19-used-cars-analysis/used-cars.html
- 2023-12-28T06:15:26.565Z
+ https://www.bryanwhiting.com/posts/2023-12-15-prompt-techniques/index.html
+ 2023-12-28T06:46:34.617Z
- https://www.bryanwhiting.com/posts/2020-12-15-data-beats-opinions/index.html
- 2023-12-28T06:15:26.569Z
+ https://www.bryanwhiting.com/posts/2022-11-12-what-is-a-model/index.html
+ 2023-12-28T06:46:34.585Z
- https://www.bryanwhiting.com/posts/2022-12-06-more-new-businesses-than-ever/index.html
- 2023-12-28T06:15:26.633Z
+ https://www.bryanwhiting.com/posts/2022-12-15-chatgpt-rshiny/index.html
+ 2023-12-28T06:46:34.597Z
- https://www.bryanwhiting.com/posts/2022-06-20-blogging-with-quarto/index.html
- 2023-12-28T06:15:26.573Z
+ https://www.bryanwhiting.com/posts/2017-02-10-ggplot-plotly/ggplot-plotly.html
+ 2023-12-28T06:46:34.505Z
- https://www.bryanwhiting.com/posts/2023-12-16-humanity-and-ai/index.html
- 2023-12-28T06:15:26.665Z
+ https://www.bryanwhiting.com/posts/2022-11-18-substance-over-style/index.html
+ 2023-12-28T06:46:34.589Z
- https://www.bryanwhiting.com/posts/2023-12-24-twas-the-night-before-christmas/index.html
- 2023-12-28T06:15:26.673Z
+ https://www.bryanwhiting.com/posts/2022-11-18-writing-a-tech-resume/index.html
+ 2023-12-28T06:46:34.589Z
- https://www.bryanwhiting.com/posts/2022-07-20-de-sciencing-data-science-and-talking-like-a-normal-person/index.html
- 2023-12-28T06:15:26.625Z
+ https://www.bryanwhiting.com/posts/2022-06-23-new-ideas-are-distractingly-exciting/index.html
+ 2023-12-28T06:46:34.545Z
- https://www.bryanwhiting.com/posts/2022-06-25-how-i-use-ticktick/index.html
- 2023-12-28T06:15:26.601Z
+ https://www.bryanwhiting.com/posts/2020-12-15-calendaring-todos/index.html
+ 2023-12-28T06:46:34.533Z
- https://www.bryanwhiting.com/posts/2023-12-15-llms-can-find-a-needle-in-the-haystack/index.html
- 2023-12-28T06:15:26.653Z
+ https://www.bryanwhiting.com/posts/2016-11-07-starting-with-jekyll/starting-with-jekyll.html
+ 2023-12-28T06:46:34.505Z
- https://www.bryanwhiting.com/gratitude/2023-12-25-grateful-for-doctors/index.html
- 2023-12-28T06:15:26.525Z
+ https://www.bryanwhiting.com/posts/2021-07-17-authors-angle-matters/index.html
+ 2023-12-28T06:46:34.533Z
+
+
+ https://www.bryanwhiting.com/gratitude/2023-12-26-i'm-grateful-for-blankets/index.html
+ 2023-12-28T06:46:34.469Zhttps://www.bryanwhiting.com/gratitude/2023-12-24-light/index.html
- 2023-12-28T06:15:26.525Z
+ 2023-12-28T06:46:34.469Zhttps://www.bryanwhiting.com/lists.html
- 2023-12-28T06:15:26.533Z
+ 2023-12-28T06:46:34.493Z
- https://www.bryanwhiting.com/christianity/2023-12-24-the-christ-child/index.html
- 2023-12-28T06:15:26.337Z
+ https://www.bryanwhiting.com/christianity/christmas-empty-tomb/index.html
+ 2023-12-28T06:46:34.281Z
- https://www.bryanwhiting.com/christianity/2023-12-13-three-little-pigs/index.html
- 2023-12-28T06:15:26.333Z
+ https://www.bryanwhiting.com/christianity/unknown-god/index.html
+ 2023-12-28T06:46:34.285Z
- https://www.bryanwhiting.com/about.html
- 2023-12-28T06:15:26.317Z
+ https://www.bryanwhiting.com/christianity/2023-12-17-christmas-celebrates-a-living-christ/index.html
+ 2023-12-28T06:46:34.277Z
- https://www.bryanwhiting.com/lists/ai/index.html
- 2023-12-28T06:15:26.537Z
+ https://www.bryanwhiting.com/christianity/2023-12-24-he-is-the-light-that-we-should-share/index.html
+ 2023-12-28T06:46:34.277Z
- https://www.bryanwhiting.com/book-reviews/2022-06-24-getting-things-done/index.html
- 2023-12-28T06:15:26.317Z
+ https://www.bryanwhiting.com/books.html
+ 2023-12-28T06:46:34.273Z
- https://www.bryanwhiting.com/book-reviews/2023-12-18-hitchhiker's-guide-to-the-galaxy/index.html
- 2023-12-28T06:15:26.321Z
+ https://www.bryanwhiting.com/book-reviews/2023-12-20-the-little-red-hen/index.html
+ 2023-12-28T06:46:34.265Z
- https://www.bryanwhiting.com/book-reviews/2023-12-23-dune/index.html
- 2023-12-28T06:15:26.329Z
+ https://www.bryanwhiting.com/book-reviews/2023-12-22-the-killers-of-the-flower-moon/index.html
+ 2023-12-28T06:46:34.269Z
- https://www.bryanwhiting.com/book-reviews/2023-12-19-101-essays-to-change-the-way-you-think-by-briana-weist/index.html
- 2023-12-28T06:15:26.325Z
+ https://www.bryanwhiting.com/book-reviews/2023-12-23-the-best-of-edgar-allen-poe/index.html
+ 2023-12-28T06:46:34.273Zhttps://www.bryanwhiting.com/book-reviews/2023-12-12-scientific-freedom/index.html
- 2023-12-28T06:15:26.321Z
+ 2023-12-28T06:46:34.261Z
- https://www.bryanwhiting.com/book-reviews/2023-12-23-the-best-of-edgar-allen-poe/index.html
- 2023-12-28T06:15:26.333Z
+ https://www.bryanwhiting.com/book-reviews/2023-12-19-101-essays-to-change-the-way-you-think-by-briana-weist/index.html
+ 2023-12-28T06:46:34.265Z
- https://www.bryanwhiting.com/book-reviews/2023-12-22-the-killers-of-the-flower-moon/index.html
- 2023-12-28T06:15:26.325Z
+ https://www.bryanwhiting.com/book-reviews/2023-12-23-dune/index.html
+ 2023-12-28T06:46:34.269Z
- https://www.bryanwhiting.com/book-reviews/2023-12-20-the-little-red-hen/index.html
- 2023-12-28T06:15:26.325Z
+ https://www.bryanwhiting.com/book-reviews/2023-12-18-hitchhiker's-guide-to-the-galaxy/index.html
+ 2023-12-28T06:46:34.261Z
- https://www.bryanwhiting.com/books.html
- 2023-12-28T06:15:26.333Z
+ https://www.bryanwhiting.com/book-reviews/2022-06-24-getting-things-done/index.html
+ 2023-12-28T06:46:34.261Z
- https://www.bryanwhiting.com/christianity/2023-12-24-he-is-the-light-that-we-should-share/index.html
- 2023-12-28T06:15:26.337Z
+ https://www.bryanwhiting.com/lists/ai/index.html
+ 2023-12-28T06:46:34.497Z
- https://www.bryanwhiting.com/christianity/2023-12-17-christmas-celebrates-a-living-christ/index.html
- 2023-12-28T06:15:26.333Z
+ https://www.bryanwhiting.com/about.html
+ 2023-12-28T06:46:34.257Z
- https://www.bryanwhiting.com/christianity/unknown-god/index.html
- 2023-12-28T06:15:26.341Z
+ https://www.bryanwhiting.com/christianity/2023-12-13-three-little-pigs/index.html
+ 2023-12-28T06:46:34.273Z
- https://www.bryanwhiting.com/christianity/christmas-empty-tomb/index.html
- 2023-12-28T06:15:26.341Z
+ https://www.bryanwhiting.com/christianity/2023-12-24-the-christ-child/index.html
+ 2023-12-28T06:46:34.277Z
+
+
+ https://www.bryanwhiting.com/img/little-lamb.png.html
+ 2023-12-28T06:46:34.485Zhttps://www.bryanwhiting.com/index.html
- 2023-12-28T06:15:26.533Z
+ 2023-12-28T06:46:34.493Z
- https://www.bryanwhiting.com/gratitude/2023-12-26-i'm-grateful-for-blankets/index.html
- 2023-12-28T06:15:26.525Z
+ https://www.bryanwhiting.com/gratitude/grateful-for-artists.html
+ 2023-12-28T06:46:34.473Z
- https://www.bryanwhiting.com/posts/2021-07-17-authors-angle-matters/index.html
- 2023-12-28T06:15:26.569Z
+ https://www.bryanwhiting.com/gratitude/2023-12-25-grateful-for-doctors/index.html
+ 2023-12-28T06:46:34.469Z
- https://www.bryanwhiting.com/posts/2016-11-07-starting-with-jekyll/starting-with-jekyll.html
- 2023-12-28T06:15:26.545Z
+ https://www.bryanwhiting.com/posts/2023-12-15-llms-can-find-a-needle-in-the-haystack/index.html
+ 2023-12-28T06:46:34.617Z
- https://www.bryanwhiting.com/posts/2020-12-15-calendaring-todos/index.html
- 2023-12-28T06:15:26.569Z
+ https://www.bryanwhiting.com/posts/2022-06-25-how-i-use-ticktick/index.html
+ 2023-12-28T06:46:34.561Z
- https://www.bryanwhiting.com/posts/2022-06-23-new-ideas-are-distractingly-exciting/index.html
- 2023-12-28T06:15:26.585Z
+ https://www.bryanwhiting.com/posts/2022-07-20-de-sciencing-data-science-and-talking-like-a-normal-person/index.html
+ 2023-12-28T06:46:34.585Z
- https://www.bryanwhiting.com/posts/2022-11-18-writing-a-tech-resume/index.html
- 2023-12-28T06:15:26.625Z
+ https://www.bryanwhiting.com/posts/2023-12-24-twas-the-night-before-christmas/index.html
+ 2023-12-28T06:46:34.633Z
- https://www.bryanwhiting.com/posts/2022-11-18-substance-over-style/index.html
- 2023-12-28T06:15:26.625Z
+ https://www.bryanwhiting.com/posts/2023-12-16-humanity-and-ai/index.html
+ 2023-12-28T06:46:34.625Z
- https://www.bryanwhiting.com/posts/2017-02-10-ggplot-plotly/ggplot-plotly.html
- 2023-12-28T06:15:26.545Z
+ https://www.bryanwhiting.com/posts/2022-06-20-blogging-with-quarto/index.html
+ 2023-12-28T06:46:34.533Z
- https://www.bryanwhiting.com/posts/2022-12-15-chatgpt-rshiny/index.html
- 2023-12-28T06:15:26.633Z
+ https://www.bryanwhiting.com/posts/2022-12-06-more-new-businesses-than-ever/index.html
+ 2023-12-28T06:46:34.593Z
- https://www.bryanwhiting.com/posts/2022-11-12-what-is-a-model/index.html
- 2023-12-28T06:15:26.625Z
+ https://www.bryanwhiting.com/posts/2020-12-15-data-beats-opinions/index.html
+ 2023-12-28T06:46:34.533Z
- https://www.bryanwhiting.com/posts/2023-12-15-prompt-techniques/index.html
- 2023-12-28T06:15:26.657Z
+ https://www.bryanwhiting.com/posts/2017-10-19-used-cars-analysis/used-cars.html
+ 2023-12-28T06:46:34.525Z
- https://www.bryanwhiting.com/posts/2018-07-02-semantic-versioning-for-data-science-models/semantic-versioning-for-data-science-models.html
- 2023-12-28T06:15:26.569Z
+ https://www.bryanwhiting.com/posts/2023-12-21-github-copilot-levels-up/index.html
+ 2023-12-28T06:46:34.629Z
- https://www.bryanwhiting.com/posts/2020-12-18-productivity-notes/index.html
- 2023-12-28T06:15:26.569Z
+ https://www.bryanwhiting.com/posts/2022-06-23-the-best-todolist-manager/index.html
+ 2023-12-28T06:46:34.549Z
- https://www.bryanwhiting.com/posts/2023-12-12-future-of-coding/index.html
- 2023-12-28T06:15:26.645Z
+ https://www.bryanwhiting.com/posts/2023-12-13-learn-by-doing/index.html
+ 2023-12-28T06:46:34.609Z
- https://www.bryanwhiting.com/posts/quarto-from-phone/index.html
- 2023-12-28T06:15:26.681Z
+ https://www.bryanwhiting.com/posts/2023-12-25-ai-robots-are-mastering-tasks-faster-than-humans-/index.html
+ 2023-12-28T06:46:34.637Z
- https://www.bryanwhiting.com/posts/2023-12-21-the-dodo/index.html
- 2023-12-28T06:15:26.673Z
+ https://www.bryanwhiting.com/posts/2022-07-03-when-data-collection-goes-too-far/index.html
+ 2023-12-28T06:46:34.581Z
- https://www.bryanwhiting.com/posts/writing-is-enlivening/index.html
- 2023-12-28T06:15:26.685Z
+ https://www.bryanwhiting.com/posts/2023-12-19-how-to-encourage-your-kids-/index.html
+ 2023-12-28T06:46:34.625Z
- https://www.bryanwhiting.com/posts/2022-12-23-who-do-you-want-to-be-known-for/index.html
- 2023-12-28T06:15:26.641Z
+ https://www.bryanwhiting.com/posts/2020-12-16-pretotyping-ml-solutions/index.html
+ 2023-12-28T06:46:34.533Z
- https://www.bryanwhiting.com/posts/write-despite-ai/index.html
- 2023-12-28T06:15:26.685Z
+ https://www.bryanwhiting.com/posts/2022-06-21-products-for-dads/index.html
+ 2023-12-28T06:46:34.533Z
- https://www.bryanwhiting.com/posts/2022-06-28-increase-focus-through-writing-down-everything/index.html
- 2023-12-28T06:15:26.601Z
+ https://www.bryanwhiting.com/posts/quarto-feature-requests/index.html
+ 2023-12-28T06:46:34.641Z
- https://www.bryanwhiting.com/posts/2022-06-28-pomodoro-principles/index.html
- 2023-12-28T06:15:26.601Z
+ https://www.bryanwhiting.com/posts/2022-06-22-downstream-impact/index.html
+ 2023-12-28T06:46:34.541Z
- https://www.bryanwhiting.com/posts/2022-06-21-the-power-of-party-pay-by-visible/index.html
- 2023-12-28T06:15:26.577Z
+ https://www.bryanwhiting.com/posts/2017-10-18-used-cars-scrape/used-cars-scrape.html
+ 2023-12-28T06:46:34.525Z
+
+
+ https://www.bryanwhiting.com/gratitude.html
+ 2023-12-28T06:46:34.469Zhttps://www.bryanwhiting.com/sortspec.html
- 2023-12-28T06:15:26.689Z
+ 2023-12-28T06:46:34.649Z
+
