Statistics for Machine Learning

Summary

In this chapter, you have learned about content-based and collaborative filtering techniques for recommending movies to users, either by considering the other users using cosine similarity, or matrix factorization to calculate by considering movie ratings. Computationally, content-based filtering is quicker to compute, but considers only one dimension, either other users or other similar movies. Whereas, in collaborative filtering, recommendations are provided by considering both the user and movie dimensions. All the Python implementation has been done from the first principles, as we do not have a good enough package for the same, and also, it is nice to know the basics. In R programming, we used the recommenderlab package to apply ...