This is a POC of a movie recommendation system. It uses a Content-Based approach by mapping the movie features into a 2d space using UMAP, then with a SVR it predicts the 5 star rating based on the user's previus ratings.

1. Copy the env file and change the values accordingly.

cp .env.example .env

2. Then start the services with:

make up

3. Open http://localhost:5173 in your browser.

4. Create an account

5. Rate the movies. Firstly, the system will recommend randomly. From 3 rated movies it will recommend based on your preference.

You can start the tests using

make test

You can evaluate the model by running:

make train

It uses the function cross_val_score to evaluate the model accuracy.