Python implementation

To better understand the process of creating the ensemble, as well as its merits, we will implement it in Python using decision trees. In this example, we will try to classify the MNIST dataset of handwritten digits. Although we have used the cancer dataset for classification examples up until now, it contains only two classes, while the number of examples is relatively small for effective bootstrapping. The digits dataset contains a considerable number of examples and is also more complex, as there is a total of 10 classes.

Implementation

For this example, we will use 1500 instances as the train set, and the remaining 297 as the test set. We will generate 10 bootstrap samples, and consequently 10 decision...