This recommender system uses two different models to determine the recommendations: Nearest Neighbors and Non-negative Matrix Factorization (can be selected). The models are trained on a reduced data set of movie ratings.
There are two methods available to get recommendations:
- By favorite movies: Select as many movies you like and get a recommendation for similar movies.
- By rating: Rate up to 10 arbitrarily selected movies and get a recommendation based on your rating.
This App was a weekly project at the SPICED Datascience Bootcamp from April to June 2023.
The code has been published on GitHub: https://github.com/yotkadata/movie_recommender
Try the App
The app is automatically deployed to render.com using GitHub actions. It runs at https://movie-recommender-tfw7.onrender.com/, but might be slow or sometimes not available.
To use the Streamlit app, clone the repository, install the requirements to your Python environment and run the app:
git clone https://github.com/yotkadata/movie_recommender cd movie_recommender/ pip install -r requirements.txt streamlit run app.py
This should open a page in your default browser at http://localhost:8501 that shows the app.