Restaurant recommendation system based on novel approach using k-means and naïve bayes classifiers

Abstract

Human opinion has become the most useful metric for recommendation systems in recent years. Before buying any product, we are looking for ratings. Growth in social media has provided an opportunity to know what interests like-minded people. In this paper, we present a k-means nearest neighbor and Naïve Bayes classifier-based systems for recommendation of restaurants. To make precise predictions and provide proficient recommendations, the data and method are most important factors. By thoroughly analyzing the literature, we came across the fact that the Facebook and Yelp are most successfully used datasets. While working, we realized the issues related to Yelp in searching from outcome of first search and time consumption. We chose Zomato that is most used restaurant-finding application which provides many attributes and users ratings. We have used Zomato data along with k-means and Naïve Bayes and achieved overall accuracy of 93%.