[WIP] Penalty on shap calculation for higher variance

[markdregan]

PR addresses #216

Overall objective: Add penalty to features that has high variance in underlying shap values - when computing feature importance. This will (in theory) encourage selection of features that have more coherency across CV folds.

Work tasks:

• [ ] Add shap_variance_penalty parameter to ShapRFECV.fit, compute & fit_compute
• [ ] Add shap_variance_penalty parameter to EarlyStoppingShapRFECV.fit, compute & fit_compute
• [ ] Add shap_variance_penalty to shap_helpers.calculate_shap_importance
• [ ] Add shap_variance_penalty to ShapModelInterpreter.fit, compute & fit_compute
• [ ] Add to doc strings throughout
• [ ] Run simulation demonstrating impact of shap_variance_penalty feature

Reviewers: LMK any changes / improvements that can be made.

[ReinierKoops]

Thanks for the PR. I’ll review it once it’s complete. Please have a look at the current GitHub actions run output. Will rerun it once you pushed new changes.

[markdregan]

@ReinierKoops - I seem to have based this PR off some older version of probatus. Just submitted fresh PR. Apologies. I'll delete this PR.