Recommending Restaurants: A Collaborative Filtering Approach

Abstract

Recommender systems are algorithms for suggesting relevant items to users (items being movies, books, products to buy or anything else depending on industries). By building up a Recommender System which could assist a client with deciding which restaurant one should visit. There are different factors depending on which a user settles on a choice of visiting a restaurant like the sort of food of the restaurant, the area of the restaurant, the climate, approximate cost, reputation, ratings, and so forth. So as to discover a descent machine learning model, we have attempted various collaborative filtering models to predict the ratings between restaurants and users. The algorithms we have implemented are the k-Nearest Neighbors algorithm and the multiclass SVM classification. Our assessment shows that the multiclass SVM classification method shows the best result. For rating prediction, we correlate user-based and item-based collaborative filtering methods.