Our aim is to develop a food recommendation system which helps users by recommending restaurants using data collected by yelp. Using Yelp’s dataset, we extract customer, restaurant profiles and ratings given for suggesting recommendations. In particular, we implement naïve baseline, singular value decomposition, hybrid cascade of K-nearest neighbor clustering, weighted bi-partite graph projection and its variants with clustering or multi-step random walk or both. Using Root Mean Squared Error, we then evaluate and compare the algorithms’ performances.
Data: We have used yelp data available at-https://www.yelp.com/dataset_challenge/ Data files which we have utilized for the project are business. json, review.json, user.json