pandaslearn 0.1.9

`pandaslearn` is a small wrapper on top of `scikit-learn` to automate common modeling tasks.

pandaslearn

pandaslearn is a small wrapper on top of scikit-learn to automate common modeling tasks.

• Create Trainer instance with Dataset and Model instances, __init__() in Trainer instance should populate Dataset and Model instance's logger attributes. Methods on Dataset and Model should be called after that, so that everything gets logged appropriately.

TODO

• TODO: visualization: add barcharts (plotnine)
• TODO: visualization: add histograms (plotnine)
• TODO: visualization: add scatterplots (plotnine)
• TODO: visualization: add lineplots (plotnine)
• TODO: visualization: add boxplots (plotnine)
• TODO: visualization: add violin plots (plotnine)
• TODO: visualization: add function to change theme (xkcd, ?)
• TODO: add a geo namespace (+ feature engineering, plots)
• TODO: add tests against a few standard fixtures (precompute values and test against them)
• TODO: integrate missingno package: functions to only compute/sort nullity
• TODO: integrate missingno package: plotnine functions matching missingno plot(including geo)
• TODO: integrate missingno package: timeseries nullity plots (just plot all timelines with gaps)
• TODO: pandas-profiling has a lot of useful analysis useful for ml. Integrate those (provide textual outcomes like dicts or dfs instead of plot)
• TODO: future integration targets: https://compose.alteryx.com/en/stable/
• TODO: future integration targets: https://featuretools.alteryx.com/en/stable/
• TODO: future integration targets: https://evalml.alteryx.com/en/stable/