Estimated rating based on hours played for video game recommendation

Abstract

This work presents a method to estimate ratings for video games based on the user’s playing hours. Based on these ratings, through collaborative filtering techniques, it is possible to make recommendations for video games without taking into account their popularity, solving the problem of long tail. The item-based k-NN algorithms and SVD++ are the ones that obtains the best results with the proposed estimation method, improving the original one and obtaining similar results in the rest of cases.