April 2018
Beginner to intermediate
282 pages
6h 52m
English
In this method, multiple layers of classifiers are stacked/piled up one over the other. The prediction probabilities of the first layer of classifiers are applied to train the second layer of classifiers and so on. The final result is achieved by employing a base classifier such as logistic regression. We can also use different algorithms, such as decision trees, random forest, or GBM, as a final layer classifier.
There is no out-of-the-box implementation for stacked ensembles in scikit-learn. However, we will demonstrate creating an automated function for stacked ensemble using scikit-learn's base algorithms in Chapter 4, Automated Algorithm Selection.