My first attempt at creating a simple recommender system.
This project demonstrates the capability of recommending small businesses for different NYC neighborhoods using the Yelp data.
Neighborhood names and locations are obtained from the ArcGIS book—NYC.gdb—Neighborhoods file. Only neighborhoods in Manhattan
and some neighborhoods in Brooklyn are selected for the convenience of analysis. The extracted file is called nbhds.txt.
Run the code to convert the neighborhoods text file to csv:
python file_converter.py
To install the dependencies, run:
pip install yelpapi
Run the code to parse Yelp reviews:
python y_parser.py
Run simple_analysis.ipynb to see the similarity analysis based on cosine distance and the recommendation analysis based on
the most similar neighborhood.
Run analysis_w_matrix_factorization.ipynb to see the analysis taking matrix factorization into account.
The recommended business results are illustrated in results.jpg, plotted using ArcGIS.