This project provides a recommender system using Collaborative Filtering via Cosine Similarity
The goals of this project are:
- Test various Machine Learning approaches for suitability for use as a recommender
- Implement user derived Collaboritive Filtering to predict users’ ratings on movies using all users.
- Implement item derived Collaboritive Filtering to predict users’ ratings on movies using all items.
- Predict the k most similar movies and users according to each movie and user.
- Improve the user derived approach using only top-K most similar user ratings.
- Prove that Collaboritive Filtering is a simple but effectice means of for recommending in a memory-based context.
- Show that using top-K users vs overall is more accurate.
$ python3 recommender.py
And
$ python3 proj.py
recommender.py written with:
- python3 - v3.7
- scikit-learn - v0.21.3
- matplotlib - v3.1.1
- numpy - v1.17.0
- pandas - v0.25.0
proj.py written with:
- python3 - v3.7
- scikit-learn - v0.15.2
- numpy - v1.17.0
- pandas - v0.15.2
- SciPy - 0.14.0
- Joel Lawrence - joel295
- Deepansh Singh - deepanshsingh8
This project is licensed under the GPL-3.0 License - see the LICENSE.md file for details