subtopic

The scikit-learn API and how other libraries differ

Use fit / predict / predict_proba / score on 3 sklearn models, then the same with XGBoost, LightGBM and CatBoost. Concepts: Estimator API; predict vs predict_proba; library differences Project A: predict_proba for Bad; CatBoost directly on the original categorical columns. Project B: Regressors: predict only (no probabilities); quantile/interval predictions come later.