In this chapter, we will cover the following recipes:
Modeling preference expressions
Understanding the data
Ingesting the movie review data
Finding the highest-scoring movies
Improving the movie-rating system
Measuring the distance between users in the preference space
Computing the correlation between users
Finding the best critic for a user
Predicting movie ratings for users
Collaboratively filtering item by item
Building a nonnegative matrix factorization model
Loading the entire dataset into the memory
Dumping the SVD-based model to the disk
Training the SVD-based model
Testing the SVD-based model