Introduction

In the previous chapter, you learned about regression problems where the target variable is continuous. A continuous variable can take any value between a minimum and maximum value. You learned how to train such models with TensorFlow.

In this chapter, you will look at another type of supervised learning problem called classification, where the target variable is discrete — meaning it can only take a finite number of values. In industry, you will most likely encounter such projects where variables are aggregated into groups such as product tiers, or classes of users, customers, or salary ranges. The objective of a classifier is to learn the patterns from the data and predict the right class for observation.

For instance, in the case of a loan provider, a classification model will try to predict whether a customer is most likely to default in the coming year based on their profile and financial position. This outcome can only take two possible values (yes...