14.5. Example: a restaurant dish recommendation engine

We’re now going to build a recommendation engine. The topic we’re going to apply this to is restaurant food. Say you’re sitting at home and you decide to go out to eat, but you don’t know where you should go or what you should order. This program will tell you both.

We’re first going to create the basic recommendation engine, which looks for things you haven’t yet tried. The second step is to improve our recommendations by using the SVD to reduce the feature space. We’ll then wrap up this program with a human-readable UI so that people can use it. Finally, we’ll discuss some problems with building recommendation engines.