Now, let's try supervised learning. As we discussed earlier, supervised models are designed to predict, or estimate, target variables—either a continuous value (regression problem) or a specific category (classification problem). Other types of problems (tagging, semantic segmentation, and many more)—can then be defined as one of those two. Let's glance over a suite of relatively simple models now.

One simple yet often pretty performant model is the k-nearest neighbors (KNN) algorithm. It is quite similar to k-means, as it also performs in Euclidean space.