Computer scienceData scienceMachine learningEnsemble learning

Ensemble methods

Types of ensembles

Report a typo

Match the descriptions of the ensembling strategies below with their names.

Match the items from left and right columns

Each base learner is trained on a slightly different variant of the training data. Ensemble prediction is computed as a simple average (regression) or majority vote (classification).

Base learners are trained one by one, with each subsequent model trying to improve performance of the previous ones. Ensemble prediction is computed as a weighted average (regression) or weighted voting (classification).

A number of different base learners are trained. Then, another model is built that learns to combine base models' predictions to achieve the best performance.

A number of different base learners are trained on the same training data. Ensemble prediction is computed as a simple average (regression) or majority vote (classification).

Boosting

Bagging

None of the above

Stacking

___

Create a free account to access the full topic