When many people think about machine learning or artificial intelligence, they probably first think about machine learning to solve classification problems. These are problems where we want to train a model to predict one of a finite number of distinct categories. For example, we may want to predict if a financial transaction is fraudulent or not fraudulent, or we may want to predict whether an image contains a hot dog, airplane, cat, and so on, or none of those things.
The categories that we try to predict could number from two to many hundreds or thousands. In addition, we could be making our predictions based on only a few attributes or many attributes. All of the scenarios arising from these combinations lead to a host of models with a corresponding host of assumptions, advantages, and disadvantages.
We will cover some of these models in this chapter and later...
