Hands-On Recommendation Systems with Python

Training and test data

The ratings DataFrame contains user ratings for movies that range from 1 to 5. Therefore, we can model this problem as an instance of supervised learning where we need to predict the rating, given a user and a movie. Although the ratings can take on only five discrete values, we will model this as a regression problem.

Consider a case where the true rating given by a user to a movie is 5. A classification model will not distinguish between the predicted ratings of 1 and 4. It will treat both as misclassified. However, a regression model will penalize the former more than the latter, which is the behavior we want.

As we saw in Chapter 5, Getting Started with Data Mining Techniques, one of the first steps towards building ...