Documentation needs a flowchart overview of components

[PeterDSteinberg]

Feedback today was:

• It is hard for someone with statistical modeling experience but new to Python or Elm's dependencies to get acquainted with basic capabilities and what the components are.

[PeterDSteinberg]

@gbrener This is a useful ML models flow chart from scikit-learn we can adapt, perhaps separately from the flow chart you and I worked on over the last week. The scikit-learn flowchart shows how to choose an estimator based on type of data (labeled or not) and data size.

Also paraphrasiing the top section of this tutorial (the Machine Learning Problem Setting section) would be helpful for Elm's range of ML options.

[PeterDSteinberg]

Let's also summarize these two JSON's of supported estimators (updated as necessary at the time of working on the issue). I made these by grepping the output of pytest to see which estimators/transformers passed by themselves when run through fit/predict as well as which ones failed in combinations of transformer and estimator in 2-step pipelines.

It is related to the flowchart because part of our documentation approach should be to better explain the degree of scikit-learn support (by sklearn subpackage/class) and cross-link the help of the scikit-learn estimators/transformers with examples/docs in Elm and related tools

supported_pipelines.txt supported_estimators.txt