Due to the complex nature of the program, we provide a conceptual explanation and then proceed to explain the details of the program.
The following figure depicts a conceptual view of ALS and how it factorizes the user/movie/rating matrix, which is a high-ranking order matrix to a lower order tall and skinny matrix, and a vector of latent factors: f(users) and f(movies).
Another way to think about it is that these factors can be used to place the movie in an n dimensional space that will be matched to a given recommendation for a given user. It is always desirable to view machine learning as a search query in a dimensional ...