Chapter 25: Model Ensembling and Stacking
This chapter covers advanced ensemble techniques beyond bagging and boosting, including model averaging and stacking for combining heterogeneous models.
-
Chapter 25.01: Ensembling and Model Averaging
We introduce ensembling, recap bagging and boosting, and cover simple and weighted model averaging together with greedy ensemble selection for choosing the averaging weights.
-
Chapter 25.02: Stacking
We explain stacking (stacked generalization), how cross-validated stacking avoids leakage between base models and the meta learner, how to stack multiple layers for heterogeneous ensembles, and look at AutoGluon as a practical application.