Bagging Classification Demo

Bagging for classification

Each bag trains a full-depth CART tree on a bootstrap sample. The ensemble predicts by majority vote over all bags.

Dataset

Criterion

Number of bags

Samples per bag

Points per class / pattern

red class blue class background shading = bagged ensemble prediction

0%training accuracy

—OOB error

—avg unique points per bag

0bags used

All trees are grown to maximum depth subject only to the inability to make a further split.