For this project, I tried to implement to a feature in Yelp that I wish existed. When you search a business on Yelp it will give you similar businesses beside the one that you had searched for, however this search only extends to businesses in the area. I can not search for bars that are similar to bar X from SF when I go to another place, for instance Chicago. This businesses profile and recommender is what I decided to then build which would take in up to three bars as inputs to find bars that were similar in a given locale. I decided to subset on bars as the type of bar that a person likes is very particular to them. For this project I ended up using the Yelp challenge dataset which has information on business in 10 different metropolitan locations in the US and Canada. The data being spread out over different areas was important as I wanted my recommender to be location agnostic.
Repository Structure
- Flask: Flask app that I built utilizing my model. Copy of this will be on a different repo which will be deployed on Heroku
- fletcher_MVP.md: Document describing my minimum viable product for this project.
- BarRecommendation_Project.ipynb: Jupyter notebook containing the code and graphics used to build the model for this project