InsideForestRegionClusterer

The canonical traditional supervised-region clusterer. The random forest generates branches; predict returns region cluster IDs and score returns adjusted mutual information.

from InsideForest import InsideForestRegionClusterer

model = InsideForestRegionClusterer(rf_params={"random_state": 42})
cluster_ids = model.fit_predict(X_train, y_train)
assignments = model.assign_regions(X_test)
regions = model.explain_regions(top_n=10)
quality = model.region_quality_report(X_test, y_test)

Output contract

• predict(X) returns region-cluster IDs, never target classes.
• Rows outside every selected region return -1.
• transform(X) returns a sample-by-region membership matrix.
• score(X, y) is adjusted mutual information (AMI), including -1.
• regions_, region_metrics_, labels_, and raw_regions_ expose the fitted artifacts.

forest_ and feature_importances_ describe the branch-generating forest; they are not class predictions produced by the clusterer. The old InsideForestClassifier name remains deprecated for migration only.