Hands-On Artificial Intelligence for Cybersecurity

Stacking

The stacking method owes its name to the fact that the ensemble estimator is constructed by superimposing two layers, in which the first consists of single estimators, whose predictions are forwarded to the underlying layer, in which another estimator has the task of classifying the predictions that are received.

Unlike the bagging and boosting methods, stacking can use different types of basic estimators, whose predictions can, in turn, be classified by a different type of algorithm than the previous ones.

Let's look at some examples of ensemble estimators.