Forests and Single Trees

MPSBoost forest and single-tree estimators share the native tree format, sampling semantics, and prediction aggregation logic.

Single trees

• DecisionTreeRegressor
• DecisionTreeClassifier

Single trees are useful for debugging, baselines, lightweight tasks, and forest composition.

Random Forest

• RandomForestRegressor
• RandomForestClassifier

Core semantics:

• bootstrap / sample fraction
• feature subsampling
• independent native trees
• regression mean aggregation
• classification vote/probability aggregation
• deterministic random_state
• deterministic n_jobs scheduling

ExtraTrees

• ExtraTreesRegressor
• ExtraTreesClassifier
• ExtraTreeRegressor
• ExtraTreeClassifier

ExtraTrees uses random threshold candidates and reuses native split, tree, prediction, and forest container logic.