Stacking the base learners

Given that many Machine Learning problems are quite challenging to solve for a single learning method, you ought to know how to build stronger solvers through the magic of ensemble models. Stacking is a simple yet quite powerful technique for building ensemble models. A stacked ensemble model is a meta-learner since it learns to predict the outcome not from the original data but from the predictions returned by the set of base learners.

In this exercise you will build a stacked ensemble out of the four base learners in models. The caret package has been preloaded and the control object holds the trainControl specs used to learn the base models.

Question

What are the components of an stacked ensemble model?

Possible Answers

Set of base models and their predictions.
Correlation matrix for the predictions made by the base models and a regression algorithm.
Error rates from base models and error rate of a supervisor model.
Predictions from base models and a supervisor model that learns how to best combine their predictions.

Exercise

Stacking the base learners

Instructions 1/4

Question

Possible Answers