Multiclass Classification

In this section we're going to learn about multiclass classification. We're going to start by constructing a multiclass classifier out of a collection of binary classifiers. There's actually two ways in which we can do that: by wiring them up according to the one versus all configuration or the one versus one configuration.

You'll see that a lot of binary classification metrics no longer work in multiclass classification, except for the confusion matrix. We will examine what a multiclass confusion matrix looks like, and introduce two new evaluation metrics: the macro average and the micro average.