This data science project focuses on building recommendation systems for restaurants in Pennsylvania, Florida, and Los Angeles, USA, using collaborative filtering techniques. The project aims to compare the performance of different collaborative filtering methods, including memory-based and model-based techniques. The Yelp dataset, containing information about businesses, reviews, and users, will be used to build the recommendation systems. The primary objective is to help a hypothetical restaurant catalog business enhance its customer experience and increase revenue by leveraging the insights gained from the project. The project's results can be applied to other businesses in the restaurant catalog industry, contributing to the advancement of recommendation systems and improving customer experiences.
Full project documentation can be found here.
- Inferential Statistics
- Machine Learning
- Data Visualization
- Predictive Modeling
- Collaborative Filtering
- Memory-based Filtering
- Model-based Filtering
- Python
- Postgres DB
- Pandas
- Jupyter
- PySpark
- Surprise (Python scikit for recommender systems)
- Mlflow (for Experiment tracking)
- notebooks/Model_Raw_Impl - Notebook containing the raw implementation of the item-item and user-user similarity techniques.
- notebooks/Model_Memory_Based_CF - Notebook containing the implementation of the memory-based collaborative filtering techniques.
- notebooks/Model_Model_Based_CF - Notebook containing the implementation of the model-based collaborative filtering techniques.