Let's now see how to use the methods we've described for solving a classification task on artificial datasets, which we can see in the following screenshot:

As we can see, these datasets contain two and three different classes of objects, so it makes sense to use methods for multi-class classification because such tasks appear more often in real life; they can be easily reduced to binary classification.

Classification is a supervised technique, so we usually have a training dataset, as well as new data for classification. To model this situation, we will use two datasets in our examples, one for training and one for testing. They come from the same distribution in one large dataset. However, the test set won't be used for training, therefore we can evaluate the accuracy metric and see how well models perform and generalize.