I finally got to the very last part of building a recommender engine. Starting from my post a couple of months ago on the topic, I created a very simple Flask app that serves the model created by the nearest-neighbors recommender. It’s not very exciting but at least shows how it can be done.
As I mention in the README for the project, the results are not exactly what I was hoping for but I think that beer recommendations are more complicated than “if you like this, you will like that”. I might pursue this further with a more complicated model or more data, but for now, I got what I wanted out of this: a basic understanding of what it takes to collect and clean the data, build a recommender model and serve it.