15)Dates&Times.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Working with Dates and Time in Pyhton"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from datetime import date\n",
    "from datetime import timedelta"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<hr>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Chapter 1: Dates and Calendars"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "bike = pd.read_csv(\"databases/capital-onebike.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[datetime.date(2016, 10, 7), datetime.date(2017, 6, 21)]"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from datetime import date\n",
    "\n",
    "#First year, then month, then day (from largest to smallest)\n",
    "two_hurricanes_dates = [date(2016, 10, 7), date(2017, 6, 21)]\n",
    "two_hurricanes_dates"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To access individual components of a date"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2016\n",
      "10\n",
      "7\n"
     ]
    }
   ],
   "source": [
    "two_hurricanes_dates = [date(2016, 10, 7), date(2017, 6, 21)]\n",
    "\n",
    "print(two_hurricanes_dates[0].year)\n",
    "print(two_hurricanes_dates[0].month)\n",
    "print(two_hurricanes_dates[0].day)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Finding the weekday of a date"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4\n"
     ]
    }
   ],
   "source": [
    "print(two_hurricanes_dates[0].weekday())\n",
    "\n",
    "# 0 = Monday, 6 = Sunday"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Math with dates"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-11-05\n"
     ]
    }
   ],
   "source": [
    "from datetime import date\n",
    "\n",
    "d1 = date(2017, 11, 5)\n",
    "d2 = date(2017, 12, 4)\n",
    "l = [d1, d2]\n",
    "\n",
    "print(min(l))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "29\n"
     ]
    }
   ],
   "source": [
    "delta = d2-d1\n",
    "\n",
    "#To obtain the number of days between two dates\n",
    "print(delta.days)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-12-04\n"
     ]
    }
   ],
   "source": [
    "from datetime import timedelta\n",
    "\n",
    "td = timedelta(days=29)\n",
    "print(d1+td)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Turning dates into strings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-11-05\n"
     ]
    }
   ],
   "source": [
    "d = date(2017, 11, 5)\n",
    "# ISO format: YYYY-MM-DD\n",
    "print(d)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['2017-11-05']\n"
     ]
    }
   ],
   "source": [
    "#Express the date in ISO 8601 format and put it in a list\n",
    "print( [d.isoformat()] )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['1992-12-31', '2000-01-01']\n"
     ]
    }
   ],
   "source": [
    "some_dates = ['2000-01-01', '1992-12-31']\n",
    "print(sorted(some_dates))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Everu other format - `d.strftime()`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017\n",
      "2017/01/05\n"
     ]
    }
   ],
   "source": [
    "d = date(2017, 1, 5)\n",
    "\n",
    "print(d.strftime(\"%Y\"))\n",
    "print(d.strftime(\"%Y/%m/%d\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Task 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ISO: Our earliest hurricane date: 1950-08-10\n",
      "US: Our earliest hurricane date: 08/10/1950\n"
     ]
    }
   ],
   "source": [
    "# Assign the earliest date to first_date\n",
    "first_date = date(1950, 8, 10)\n",
    "\n",
    "# Convert to ISO and US formats\n",
    "iso = \"Our earliest hurricane date: \" + first_date.isoformat()\n",
    "us = \"Our earliest hurricane date: \" + first_date.strftime(\"%m/%d/%Y\")\n",
    "\n",
    "print(\"ISO: \" + iso)\n",
    "print(\"US: \" + us)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Task 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1992-08\n",
      "August (1992)\n",
      "1992-239\n"
     ]
    }
   ],
   "source": [
    "# Create a date object\n",
    "date1 = date(1992, 8, 26)\n",
    "# Print the date in the format 'YYYY-MM'\n",
    "print(date1.strftime(\"%Y-%m\"))\n",
    "\n",
    "\n",
    "# Print the date in the format 'MONTH (YYYY)'\n",
    "print(date1.strftime(\"%B (%Y)\"))\n",
    "\n",
    "\n",
    "# Print the date in the format 'YYYY-DDD'\n",
    "print(date1.strftime(\"%Y-%j\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<hr>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Chapter 2: Combining Dates and Times"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the previous chapter we used `from datetime import date`, however now we are going to use `from datetime import datetime`, which will allow us to not only specify the date, but also the time "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-10-01 15:23:25\n"
     ]
    }
   ],
   "source": [
    "from datetime import datetime \n",
    "\n",
    "dt = datetime(2017, 10, 1, 15, 23, 25)\n",
    "print(dt)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "or you can use named arguments"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-10-01 15:23:25\n"
     ]
    }
   ],
   "source": [
    "dt = datetime(year=2017, month=10, day=1,\n",
    "              hour=15, minute=23, second=25)\n",
    "print(dt)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Replacing parts of a datetime"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Old: 2017-10-01 15:23:25\n",
      "New: 2017-10-01 15:00:00\n"
     ]
    }
   ],
   "source": [
    "print(f\"Old: {dt}\")\n",
    "\n",
    "dt_hr = dt.replace(minute=0, second=0)\n",
    "print(f\"New: {dt_hr}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### `strftime` and `strptime`"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**printing and parsing datetimes**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-12-30\n",
      "2017-12-30 15:19:13\n",
      "15:19:13 on 2017/12/30\n",
      "2017-12-30T15:19:13\n"
     ]
    }
   ],
   "source": [
    "dt = datetime(2017, 12, 30, 15, 19, 13)\n",
    "print(dt.strftime(\"%Y-%m-%d\"))\n",
    "\n",
    "print(dt.strftime(\"%Y-%m-%d %H:%M:%S\"))\n",
    "\n",
    "print(dt.strftime(\"%H:%M:%S on %Y/%m/%d\"))\n",
    "\n",
    "print(dt.isoformat()) #standartized option"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the cell above all our dates are stored as str"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**parsing datetime with `strptime` (string parse time)**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'datetime.datetime'>\n",
      "2017-12-30 15:19:13\n"
     ]
    }
   ],
   "source": [
    "#The first argument is the string to turn into the datetime\n",
    "#Second argument is the format string that says how to do it\n",
    "dt = datetime.strptime(\"12/30/2017 15:19:13\",\n",
    "                       \"%m/%d/%Y %H:%M:%S\")\n",
    "print(type(dt))\n",
    "print(dt)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Summary - `strftime` converts a datetime object to a string, while `strptime` converts a string to a datetime object**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "timestamp (number of seconds from 1970...)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-12-30 23:19:13\n"
     ]
    }
   ],
   "source": [
    "#A timestamp\n",
    "ts = 1514665153.0\n",
    "\n",
    "#Convert to datetime and print\n",
    "print(datetime.fromtimestamp(ts))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<table>\n",
    "<tbody>\n",
    "    <tr>\n",
    "        <td>%Y</td>\n",
    "        <td>4 digit year (0000-9999)</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>%m</td>\n",
    "        <td>3 digit month (1-12)</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>%d</td>\n",
    "        <td>2 digit day (1-31)</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>%H</td>\n",
    "        <td>2 digit hour (0-23)</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>%M</td>\n",
    "        <td>2 digit minute (0-59)</td>\n",
    "    </tr>\n",
    "    <tr>\n",
    "        <td>%S</td>\n",
    "        <td>2 digit second (0-59)</td>\n",
    "    </tr>\n",
    "</tbody>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Task 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1986-12-15 08:00:00\n"
     ]
    }
   ],
   "source": [
    "# Import the datetime class\n",
    "from datetime import datetime\n",
    "\n",
    "# Starting string, in MM/DD/YYYY HH:MM:SS format\n",
    "s = '12/15/1986 08:00:00'\n",
    "\n",
    "# Write a format string to parse s\n",
    "fmt = '%m/%d/%Y %H:%M:%S'\n",
    "\n",
    "# Create a datetime object d\n",
    "d = datetime.strptime(s, fmt)\n",
    "\n",
    "# Print d\n",
    "print(d)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Task 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[datetime.datetime(2017, 12, 30, 23, 19, 13), datetime.datetime(2017, 12, 30, 23, 9, 3)]\n"
     ]
    }
   ],
   "source": [
    "# Import datetime\n",
    "from datetime import datetime\n",
    "\n",
    "# Starting timestamps\n",
    "timestamps = [1514665153, 1514664543]\n",
    "\n",
    "# Datetime objects\n",
    "dts = []\n",
    "\n",
    "# Loop\n",
    "for ts in timestamps:\n",
    "  dts.append(datetime.fromtimestamp(ts))\n",
    "  \n",
    "# Print results\n",
    "print(dts)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Working with durations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1450.0\n"
     ]
    }
   ],
   "source": [
    "start = datetime(2017, 10, 8, 23, 46, 47)\n",
    "end = datetime(2017, 10, 9, 0, 10, 57)\n",
    "\n",
    "duration = end - start\n",
    "#Getting total seconds\n",
    "print(duration.total_seconds())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-10-08 23:46:47\n",
      "2017-10-08 23:46:48\n"
     ]
    }
   ],
   "source": [
    "from datetime import timedelta\n",
    "\n",
    "delta1 = timedelta(seconds=1)\n",
    "print(start)\n",
    "#One second later\n",
    "print(start+delta1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-10-08 23:46:47"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "2017-10-09 23:46:48\n"
     ]
    }
   ],
   "source": [
    "delta1 = timedelta(days=1, seconds=1)\n",
    "print(start)\n",
    "#One day and one second later\n",
    "print(start+delta1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<hr>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Chapter 3: Time zones and Daylight Saving "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### UTC offsets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-12-30 15:09:03-05:00\n"
     ]
    }
   ],
   "source": [
    "from datetime import datetime, timedelta, timezone\n",
    "\n",
    "#US Eastern Standard time zone\n",
    "ET = timezone(timedelta(hours=-5))\n",
    "\n",
    "#Timezone-aware datetime\n",
    "dt = datetime(2017, 12, 30, 15, 9, 3, tzinfo=ET)\n",
    "print(dt)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, in the output above we can see that the datetime incudes UTC offset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-12-31 01:39:03+05:30\n"
     ]
    }
   ],
   "source": [
    "#India Standard time zone\n",
    "IST = timezone(timedelta(hours=5, minutes=30))\n",
    "\n",
    "#Convert to IST\n",
    "print(dt.astimezone(IST))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is another approach to add tzinfo"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-12-30 15:09:03+00:00\n"
     ]
    }
   ],
   "source": [
    "print(dt.replace(tzinfo=timezone.utc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The thing is that when we use **`dt.replace` we only change offset**, however when we use **`.astimezone` we change both offset and actual time**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Task 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-10-01T15:26:26+11:00\n"
     ]
    }
   ],
   "source": [
    "# Import datetime, timedelta, timezone\n",
    "from datetime import datetime, timedelta, timezone\n",
    "\n",
    "# Create a timezone for Australian Eastern Daylight Time, or UTC+11\n",
    "aedt = timezone(timedelta(hours=11))\n",
    "\n",
    "# October 1, 2017 at 15:26:26, UTC+1\n",
    "dt = datetime(2017, 10, 1, 15, 26, 26, tzinfo=aedt)\n",
    "\n",
    "# Print results\n",
    "print(dt.isoformat())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-10-01 04:26:26+00:00\n"
     ]
    }
   ],
   "source": [
    "#To set utc time:\n",
    "dt = dt.astimezone(timezone.utc)\n",
    "print(dt)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Time zone database"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tzfile('America/New_York')\n"
     ]
    }
   ],
   "source": [
    "from datetime import datetime \n",
    "from dateutil import tz\n",
    "\n",
    "# Eastern time\n",
    "et = tz.gettz(\"America/New_York\") #Continent/Nearest Major City\n",
    "print(et)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "No we can pass in this tz of the USA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-12-30 15:19:03-05:00\n",
      "2017-10-01 15:23:25-04:00\n"
     ]
    }
   ],
   "source": [
    "#Notice the different change in the UTC depending on the time of the year\n",
    "\n",
    "last = datetime(2017, 12, 30, 15, 19, 3, tzinfo=et)\n",
    "print(last)\n",
    "\n",
    "first = datetime(2017, 10, 1, 15, 23, 25, tzinfo=et)\n",
    "print(first)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Starting Daylight Saving Time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'2017-03-12T01:59:59'"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "spring_ahead_159am = datetime(2017, 3, 12, 1, 59, 59)\n",
    "spring_ahead_159am.isoformat()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'2017-03-12T03:00:00'"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "spring_ahead_3am = datetime(2017, 3, 12, 3, 0, 0)\n",
    "spring_ahead_3am.isoformat()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3601.0"
      ]
     },
     "execution_count": 99,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(spring_ahead_3am - spring_ahead_159am).total_seconds()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "However the output above is wrong as it should be only 1 seconds due to changing the clocks one hour forward at that time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [],
   "source": [
    "from datetime import timezone, timedelta\n",
    "\n",
    "EST = timezone(timedelta(hours=-5))\n",
    "EDT = timezone(timedelta(hours=-4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-03-12T01:59:59-05:00\n",
      "2017-03-12T03:00:00-04:00\n"
     ]
    }
   ],
   "source": [
    "spring_ahead_159am = spring_ahead_159am.replace(tzinfo=EST)\n",
    "print(spring_ahead_159am.isoformat())\n",
    "\n",
    "spring_ahead_3am = spring_ahead_3am.replace(tzinfo=EDT)\n",
    "print(spring_ahead_3am.isoformat())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0"
      ]
     },
     "execution_count": 102,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(spring_ahead_3am - spring_ahead_159am).total_seconds()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And now we got the correct answer by setting the correct difference between local and UTC time"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Another way to avoid dealing with this issue is to use tz from the start"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [],
   "source": [
    "from dateutil import tz\n",
    "\n",
    "eastern = tz.gettz(\"America/New_York\")\n",
    "\n",
    "spring_ahead_159am = datetime(2017, 3, 12, 1, 59, 59,\n",
    "                              tzinfo = eastern)\n",
    "\n",
    "spring_ahead_3am = datetime(2017, 3, 12, 3, 0, 0,\n",
    "                            tzinfo = eastern)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Task 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2017-03-12T00:00:00-05:00 to 2017-03-12T06:00:00-04:00\n",
      "6.0\n",
      "5.0\n"
     ]
    }
   ],
   "source": [
    "# Import datetime, timedelta, tz, timezone\n",
    "from datetime import datetime, timedelta, timezone\n",
    "from dateutil import tz\n",
    "\n",
    "# Start on March 12, 2017, midnight, then add 6 hours\n",
    "start = datetime(2017, 3, 12, tzinfo = tz.gettz('America/New_York'))\n",
    "end = start + timedelta(hours=6)\n",
    "print(start.isoformat() + \" to \" + end.isoformat())\n",
    "\n",
    "# How many hours have elapsed?\n",
    "print((end - start).total_seconds()/(60*60))\n",
    "\n",
    "# What if we move to UTC?\n",
    "print((end.astimezone(timezone.utc) - start.astimezone(timezone.utc))\\\n",
    "      .total_seconds()/(60*60))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ending Daylight Saving Time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 105,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "eastern = tz.gettz(\"US/Eastern\")\n",
    "\n",
    "first_1am = datetime(2017, 11, 5, 1, 0, 0,\n",
    "                     tzinfo = eastern)\n",
    "tz.datetime_ambiguous(first_1am)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We see that this time is indeed ambiguous as it could occur at two differenct UTC times"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {},
   "outputs": [],
   "source": [
    "eastern = tz.gettz(\"US/Eastern\")\n",
    "\n",
    "second_1am = datetime(2017, 11, 5, 1, 0, 0,\n",
    "                     tzinfo = eastern)\n",
    "second_1am = tz.enfold(second_1am)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0"
      ]
     },
     "execution_count": 107,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(first_1am - second_1am).total_seconds()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3600.0"
      ]
     },
     "execution_count": 108,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "first_1am = first_1am.astimezone(tz.UTC)\n",
    "second_1am = second_1am.astimezone(tz.UTC)\n",
    "\n",
    "(second_1am - first_1am).total_seconds()\n",
    "#1 hour apart"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Overall when we are unsure about some dates and times the best approach is to convert them to UTC"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<hr>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Chapter 4: Dates and Times in Pandas"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Reading date and time data in Pandas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Start date</th>\n",
       "      <th>End date</th>\n",
       "      <th>Start station number</th>\n",
       "      <th>Start station</th>\n",
       "      <th>End station number</th>\n",
       "      <th>End station</th>\n",
       "      <th>Bike number</th>\n",
       "      <th>Member type</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2017-10-01 15:23:25</td>\n",
       "      <td>2017-10-01 15:26:26</td>\n",
       "      <td>31038</td>\n",
       "      <td>Glebe Rd &amp; 11th St N</td>\n",
       "      <td>31036</td>\n",
       "      <td>George Mason Dr &amp; Wilson Blvd</td>\n",
       "      <td>W20529</td>\n",
       "      <td>Member</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2017-10-01 15:42:57</td>\n",
       "      <td>2017-10-01 17:49:59</td>\n",
       "      <td>31036</td>\n",
       "      <td>George Mason Dr &amp; Wilson Blvd</td>\n",
       "      <td>31036</td>\n",
       "      <td>George Mason Dr &amp; Wilson Blvd</td>\n",
       "      <td>W20529</td>\n",
       "      <td>Casual</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2017-10-02 06:37:10</td>\n",
       "      <td>2017-10-02 06:42:53</td>\n",
       "      <td>31036</td>\n",
       "      <td>George Mason Dr &amp; Wilson Blvd</td>\n",
       "      <td>31037</td>\n",
       "      <td>Ballston Metro / N Stuart &amp; 9th St N</td>\n",
       "      <td>W20529</td>\n",
       "      <td>Member</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           Start date            End date  Start station number  \\\n",
       "0 2017-10-01 15:23:25 2017-10-01 15:26:26                 31038   \n",
       "1 2017-10-01 15:42:57 2017-10-01 17:49:59                 31036   \n",
       "2 2017-10-02 06:37:10 2017-10-02 06:42:53                 31036   \n",
       "\n",
       "                   Start station  End station number  \\\n",
       "0           Glebe Rd & 11th St N               31036   \n",
       "1  George Mason Dr & Wilson Blvd               31036   \n",
       "2  George Mason Dr & Wilson Blvd               31037   \n",
       "\n",
       "                            End station Bike number Member type  \n",
       "0         George Mason Dr & Wilson Blvd      W20529      Member  \n",
       "1         George Mason Dr & Wilson Blvd      W20529      Casual  \n",
       "2  Ballston Metro / N Stuart & 9th St N      W20529      Member  "
      ]
     },
     "execution_count": 109,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "#if we want pandas to treat some columns as datetimes\n",
    "rides = pd.read_csv(\"databases/capital-onebike.csv\",\n",
    "                    parse_dates = [\"Start date\", \"End date\"])\n",
    "rides.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Start date                               2017-10-02 06:37:10\n",
       "End date                                 2017-10-02 06:42:53\n",
       "Start station number                                   31036\n",
       "Start station                  George Mason Dr & Wilson Blvd\n",
       "End station number                                     31037\n",
       "End station             Ballston Metro / N Stuart & 9th St N\n",
       "Bike number                                           W20529\n",
       "Member type                                           Member\n",
       "Name: 2, dtype: object"
      ]
     },
     "execution_count": 110,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#can get a particular row (number 2)\n",
    "rides.iloc[2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In case you want to specify the format of a datetime manually you should use `pd.to_datetime()`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [],
   "source": [
    "rides[\"Start date\"] = pd.to_datetime(rides[\"Start date\"],\n",
    "                                    format = \"%Y-%m-%d %H:%M:%S\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Timestamp('2017-10-02 06:37:10')"
      ]
     },
     "execution_count": 112,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rides[\"Start date\"].iloc[2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0   0 days 00:03:01\n",
      "1   0 days 02:07:02\n",
      "2   0 days 00:05:43\n",
      "3   0 days 00:21:18\n",
      "4   0 days 00:21:17\n",
      "Name: Duration, dtype: timedelta64[ns]\n"
     ]
    }
   ],
   "source": [
    "rides[\"Duration\"] = rides[\"End date\"] - rides[\"Start date\"]\n",
    "\n",
    "print(rides[\"Duration\"].head())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We use a backslash for readability"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0     181.0\n",
       "1    7622.0\n",
       "2     343.0\n",
       "3    1278.0\n",
       "4    1277.0\n",
       "Name: Duration, dtype: float64"
      ]
     },
     "execution_count": 114,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rides[\"Duration\"]\\\n",
    "    .dt.total_seconds()\\\n",
    "    .head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Summarizing datetime data in Pandas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Timedelta('0 days 00:19:38.931034482')"
      ]
     },
     "execution_count": 115,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rides[\"Duration\"].mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Timedelta('3 days 22:58:10')"
      ]
     },
     "execution_count": 116,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rides[\"Duration\"].sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.04348417785917786"
      ]
     },
     "execution_count": 117,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rides[\"Duration\"].sum() / timedelta(days=91)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Member type\n",
       "Member    0.813793\n",
       "Casual    0.186207\n",
       "Name: proportion, dtype: float64"
      ]
     },
     "execution_count": 118,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rides[\"Member type\"].value_counts(normalize=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Member type\n",
       "Casual    1994.666667\n",
       "Member     992.279661\n",
       "Name: Duration seconds, dtype: float64"
      ]
     },
     "execution_count": 119,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rides[\"Duration seconds\"] = rides[\"Duration\"].dt.total_seconds()\n",
    "\n",
    "rides.groupby(\"Member type\")[\"Duration seconds\"].mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`.resample` takes a unit of time sush as month and a datetime column to group on"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the cell below we calculate the average duration in seconds for every month"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Start date\n",
       "2017-10-31    1886.453704\n",
       "2017-11-30     854.174757\n",
       "2017-12-31     635.101266\n",
       "Freq: ME, Name: Duration seconds, dtype: float64"
      ]
     },
     "execution_count": 120,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#ME here stands for month\n",
    "rides.resample(\"ME\", on = \"Start date\")[\"Duration seconds\"].mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Member type\n",
       "Casual     54\n",
       "Member    236\n",
       "dtype: int64"
      ]
     },
     "execution_count": 121,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rides.groupby(\"Member type\").size()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To add a plot, just add a plot method at the end of your code like this"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Axes: xlabel='Start date'>"
      ]
     },
     "execution_count": 122,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "rides\\\n",
    "    .resample(\"D\", on = \"Start date\")\\\n",
    "    [\"Duration seconds\"]\\\n",
    "    .mean()\\\n",
    "    .plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see an oulier which is likely a bike repair"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Task 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Start date  Member type\n",
      "2017-10-31  Member         0.768519\n",
      "            Casual         0.231481\n",
      "2017-11-30  Member         0.825243\n",
      "            Casual         0.174757\n",
      "2017-12-31  Member         0.860759\n",
      "            Casual         0.139241\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "# Resample rides to be monthly on the basis of Start date\n",
    "monthly_rides = rides.resample(\"ME\", on=\"Start date\")['Member type']\n",
    "\n",
    "# Take the ratio of the .value_counts() over the total number of rides\n",
    "print(monthly_rides.value_counts() / monthly_rides.size())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Task 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Member type  Start date\n",
      "Casual       2017-10-31          0 days 00:27:16\n",
      "             2017-11-30   0 days 00:19:19.500000\n",
      "             2017-12-31          0 days 00:14:10\n",
      "Member       2017-10-31          0 days 00:11:11\n",
      "             2017-11-30          0 days 00:10:55\n",
      "             2017-12-31   0 days 00:06:27.500000\n",
      "Name: Duration, dtype: timedelta64[ns]\n"
     ]
    }
   ],
   "source": [
    "# Group rides by member type, and resample to the month\n",
    "grouped = rides.groupby('Member type')\\\n",
    "  .resample(\"ME\", on = \"Start date\")\n",
    "\n",
    "# Print the median duration for each group\n",
    "print(grouped[\"Duration\"].median())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Additional datetime methods in Pandas\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-3346.0"
      ]
     },
     "execution_count": 125,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rides[\"Duration\"].dt.total_seconds().min()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We see that smth is wrong"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Try te set a timezone...\n",
    "rides[\"Start date\"] = rides[\"Start date\"]\\\n",
    "    .dt.tz_localize(\"America/New_York\")\n",
    "\n",
    "#We will get an error as we have an ambiguous value"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {},
   "outputs": [],
   "source": [
    "rides[\"Start date\"] = rides[\"Start date\"]\\\n",
    "    .dt.tz_localize(\"America/New_York\", ambiguous=\"NaT\")\n",
    "#Means that bad results sould be set as not a time\n",
    "\n",
    "rides[\"End date\"] = rides[\"End date\"]\\\n",
    "    .dt.tz_localize(\"America/New_York\", ambiguous=\"NaT\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "116.0"
      ]
     },
     "execution_count": 128,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rides[\"Duration\"] = rides[\"End date\"] - rides[\"Start date\"]\n",
    "\n",
    "rides[\"Duration\"].dt.total_seconds().min()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Start date                        NaT\n",
       "End date                          NaT\n",
       "Start station number            31615\n",
       "Start station           6th & H St NE\n",
       "End station number              31627\n",
       "End station             3rd & M St NE\n",
       "Bike number                    W20529\n",
       "Member type                    Member\n",
       "Duration                          NaT\n",
       "Duration seconds              -3346.0\n",
       "Name: 129, dtype: object"
      ]
     },
     "execution_count": 129,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rides.iloc[129]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Other datetime operations in Pandas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    2017\n",
       "1    2017\n",
       "2    2017\n",
       "Name: Start date, dtype: int32"
      ]
     },
     "execution_count": 130,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rides[\"Start date\"]\\\n",
    "    .head(3)\\\n",
    "    .dt.year"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    Sunday\n",
       "1    Sunday\n",
       "2    Monday\n",
       "Name: Start date, dtype: object"
      ]
     },
     "execution_count": 131,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rides[\"Start date\"]\\\n",
    "    .head(3)\\\n",
    "    .dt.day_name()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We also have an ability to move rows"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0                         NaT\n",
       "1   2017-10-01 15:26:26-04:00\n",
       "2   2017-10-01 17:49:59-04:00\n",
       "3   2017-10-02 06:42:53-04:00\n",
       "4   2017-10-02 09:18:03-04:00\n",
       "Name: End date, dtype: datetime64[ns, America/New_York]"
      ]
     },
     "execution_count": 132,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rides[\"End date\"].shift(1).head()\n",
    "\n",
    "#Below we have moved the first row from being at index 0 to index1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Task 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Localize the Start date column to America/New_York\n",
    "rides['Start date'] = rides['Start date'].dt.tz_localize('America/New_York', \n",
    "                                \t\t\t\t\t\t ambiguous='NaT')\n",
    "\n",
    "# Print first value\n",
    "print(rides['Start date'].iloc[0])\n",
    "\n",
    "# Convert the Start date column to Europe/London\n",
    "rides['Start date'] = rides['Start date'].dt.tz_convert(\"Europe/London\")\n",
    "\n",
    "# Print the new value\n",
    "print(rides['Start date'].iloc[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Task 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 135,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ride start weekday\n",
      "Friday      0 days 00:12:04.500000\n",
      "Monday      0 days 00:13:30.500000\n",
      "Saturday           0 days 00:07:42\n",
      "Sunday             0 days 00:15:17\n",
      "Thursday           0 days 00:10:52\n",
      "Tuesday     0 days 00:10:41.500000\n",
      "Wednesday          0 days 00:09:45\n",
      "Name: Duration, dtype: timedelta64[ns]\n"
     ]
    }
   ],
   "source": [
    "# Add a column for the weekday of the start of the ride\n",
    "rides['Ride start weekday'] = rides['Start date'].dt.day_name()\n",
    "\n",
    "# Print the median trip time per weekday\n",
    "print(rides.groupby(\"Ride start weekday\")['Duration'].median())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A nice recap in the last wrap-up video - check it out"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}