TSML Primer: Add article series about "Machine Learning for Time Series Data" #54
Conversation
- Expand canonical "time series" entry-point page - Add dedicated time series sub-pages about: - Time Series Basics - Advanced Time Series Analysis - Connectivity Options - Video Tutorials - Use "time series" 2-gram everywhere - Improve page about "Industrial Data" - Improve page about "Document Store" - ML: Add section about "Exploratory data analysis (EDA)"
We need to limit sphinx-autobuild to a previous release. The new one has raises build errors on our docs.
amotl
force-pushed
the
amo/timeseries-primer
branch
from
d6cb363
to
9bb88db
Compare
amotl
changed the title
amotl
changed the title
Co-authored-by: Andreas Motl <[email protected]> Co-authored-by: ckurze <[email protected]>
- timeseries/basic - ml/index
- Use Intersphinx links where possible - Use cratedb.com instead of crate.io
HTTPSConnectionPool(host='aka.ms', port=443): Read timed out.
amotl
force-pushed
the
amo/timeseries-primer
branch
from
9bb88db
to
41112dd
Compare
amotl
changed the title
amotl
added
enhancement
Add link to primer on "Advanced Time Series Analysis" page.
amotl
requested review from
hammerhead,
surister,
hlcianfagna and
ckurze
and removed request for
hammerhead and
hlcianfagna
There was a problem hiding this comment.
Just adding an inline comment at a relevant spot to invite you to discuss further refactoring work on this piece. Please use it to discuss high-level details like shape and structure, and create dedicated conversations on relevant other spots to discuss corresponding details there. Thanks!
| (timeseries-ml-primer)= | ||
| (tsml-primer)= | ||
|
|
||
| # Machine Learning for Time Series Data | ||
|
|
||
| A primer about how to apply machine learning procedures to time series data. | ||
|
|
||
|
|
||
| ## Time Series Modeling | ||
|
|
||
| This section will introduce you to the concept of time series modeling, and | ||
| discusses the main obstacles usually faced with during its implementation in | ||
| production. | ||
|
|
||
| ```{toctree} | ||
| :maxdepth: 2 | ||
|
|
||
| introduction | ||
| anomaly-detection | ||
| ``` | ||
|
|
||
|
|
||
| ## MLOps | ||
|
|
||
| MLOps / ML Ops is a paradigm that aims to deploy and maintain machine learning | ||
| models in production reliably and efficiently, including experiment tracking. | ||
|
|
||
| ```{toctree} | ||
| :maxdepth: 2 | ||
|
|
||
| mlops-intro | ||
| mlops-cratedb-mlflow | ||
| mlops-cratedb-sql | ||
| ``` |
There was a problem hiding this comment.
RFC
Let's discuss the guidance into this content section, as reflected here. Let us know what you think about it, and any other ideas which come to mind when thinking about reshaping and improving this whole content section further. Thanks!
Status Quo
People who have been involved in the original creation process can see that we've already split the whole big behemoth into two subsections now, shuffled across five pages, intended to provide an improved bundling
- Introduction + Time Series Modeling
vs. - MLOps
... in order to separate concerns. Beforehand, the document was conceived and intended to be published as three items of a blog post series. 1
Another step in the refactoring process so far, already implemented on behalf of 41112dd, was to also decrease total page lengths, by breaking out all the hands-on, code-focused sections into dedicated pages, that's why we have five now 2.
So far, we haven't added any guiding style elements, like the tiles seen on other pages, yet. Shaping the structure and layout before adding priming and color makes sense.
Footnotes
-
We believe that this has been cancelled, because publishing technical content on HubSpot does not work well, even when using https://github.com/crate-workbench/hubspot-tech-writing. ↩
-
On the matter of further improvements, please also see https://github.com/crate/cratedb-examples/issues/341#issuecomment-1971228147. ↩
amotl
force-pushed
the
amo/timeseries-more
branch
2 times, most recently
from
9bcbb04
to
2a790a8
Compare
amotl
force-pushed
the
amo/timeseries-more
branch
6 times, most recently
from
9d205d0
to
6669b3a
Compare
amotl
force-pushed
the
amo/timeseries-more
branch
4 times, most recently
from
b0dd301
to
9e16cd8
Compare
amotl
force-pushed
the
amo/timeseries-more
branch
from
9e16cd8
to
07ece54
Compare
amotl
force-pushed
the
amo/timeseries-more
branch
2 times, most recently
from
e5ad55e
to
533fda0
Compare
amotl
force-pushed
the
amo/timeseries-more
branch
2 times, most recently
from
a14ef85
to
d718397
Compare
About
Add the primer article about »machine learning for time series data« by @andnig. Thanks!
Preview
https://cratedb-guide--54.org.readthedocs.build/domain/timeseries/ml-primer/
Backlog
A few sections of the hands-on examples have been broken out into Jupyter Notebooks already. Make sure to swap content accordingly, and reference or even embed the new content.