This project is to build a recommender to predict Yelp business ratings from active users.
- Baseline-Bias: One of our baselines to compare all the other recommenders against. It estimates the predicted rating using the biases of the user's and business's ratings.
- Baseline-CF: One of our baselines to compare all the other recommenders against. It uses a -based collaborative filtering () to predict ratings.
This project is broken down to a few components. Final results are summarized in final_recommendation_results.ipynb.
All the training was performed in model_recommendation.ipynb using algorithms in source/model_recommender.py files.
Data is stored in data dir (dataset removed due to github limitations but can be downloaded at https://www.yelp.com/dataset/challenge) and results are generated in data/results dir.
.
├── README.md
├── baseline_recommendation.ipynb
├── final_recommendation_results.ipynb
├── data
│ ├── results
│ │ ├── temp.pkl
│ │ └── temp.pkl
├── environment.yml
├── images
│ ├── temp.png
│ └── temp.png
└── source
├── baseline_recommender.py
├── cf_recommender.py
└── utils.pyInstall Spark version 2.4.4 and then set up conda env with conda env create -f environment.yml