Hands-On Ensemble Learning with Python

Bagging for classification

Scikit-learn's implementation of bagging lies in the sklearn.ensemble package. BaggingClassifier is the corresponding class for classification problems. It has a number of interesting parameters, allowing for greater flexibility. It can use any scikit-learn estimator by specifying it with base_estimator. Furthermore, n_estimators dictates the ensemble's size (and, consequently, the number of bootstrap samples that will be generated), while n_jobs dictates how many jobs (processes) will be used to train and predict with each base learner. Finally, if set to True, oob_score calculates the out-of-bag score for the base learners.

Using the actual classifier is straightforward and similar to all other scikit-learn estimators. ...