subtopic

Random Forest (classifier | regressor)

Fit a random forest; check out-of-bag score; inspect importance; tune n_estimators, max_features, min_samples_leaf. Concepts: Bagging; bootstrap samples; random feature subsets; OOB estimate Project A: Random forest vs logistic regression on cost; usually higher ROC-AUC, needs calibration for risk scores. Project B: Random forest regressor with lag/calendar features; compare peak-hour MAE.