Abstract
Recommendation systems are leveraged by large enterprises to help users discover new and relevant items to create a great experience and generate huge revenues. In today’s world, recommending products or businesses through recommendation systems have become ubiquitous, and platforms like Yelp, Amazon, YouTube, Netflix are leveraging these recommendation systems to provide the best user experience. This project mainly focuses on creating a recommendation system based on the restaurant business. People have worked on several recommendation systems in the past for the restaurants, where most of them recommend a similar restaurant to a user based on their previous visits. This project aims to develop a recommendation system that recommends a new restaurant to the user, which they had never visited before and would probably enjoy. In this project, the recommendation system was developed using the Yelp dataset. It recommends a new restaurant that a user would like by calculating two metrics: similarity of it to the previous restaurants the user visited and the rating the user would most likely give it. The rating prediction was implemented using machine learning models like Linear Regression, Random Forest, and Neural Networks, and these models were evaluated using the Root Mean Square Error (RMSE) value. Out of the three models, Neural Networks gave the best RMSE value of 0.74. The similarity between restaurants was calculated with content-based filtering method using cosine similarity. Rating prediction and similarity were combined to provide the final recommended restaurants, along with the user predicted ratings.