A package for the cranky data scientist to make cross-validation into an over-abstracted mess that may yet still end up saving on some typing.
- This package is a prototype, and breaking changes are imminent.
- No PRs will be accepted.
- This package is a prototype, and breaking changes will happen.
Try to make the following common tasks suck less:
- Column name awareness b/c columns aren't John Smiths and/or Jane Does.
- View model predictions on individual instances.
- Get model coefficients and metrics.
- Compare models side by side.
- Work with multiple 3rd party models simultaenously (e.g., statsmodels and sklearn).