14_ensemble_methods

Ensemble methods

By Gianluca Campanella (g.campanella@estimand.com)

Objectives

By the end of the session, you should be able to:

• Describe the main differences between bagging, boosting and stacking
• Explain why ensembles achieve higher accuracy than individual predictors
• Use scikit-learn and XGBoost to fit boosted models

Plan

The session is designed to be delivered over three hours (including breaks).

Topic	Time
Ensemble methods	45 minutes
Boosting using scikit-learn and XGBoost	45 minutes
Exercises	60 minutes

Materials

• Ensemble methods
• Boosting using scikit-learn and XGBoost
• Exercises and solutions