Movie Rating Prediction with Matrix Factorization Algorithm

Abstract

Recommendation systems are one of the research areas studied intensively in the last decades and several solutions have been elicited for problems in different domains for recommending. Recommendation may differ as content, collaborative filtering or both. Other than known challenges in collaborative filtering techniques, accuracy and computational cost at a large scale data still at saliency. In this paper we proposed an approach by utilizing matrix value factorization for predicting rating i by user j with the sub matrix as k-most similar items specific to user i for all users who rated them all. In an attempt, previously predicted values are used for subsequent predictions and we have investigated the accuracy of neighborhood methods by applying our method on Netflix Prize (http://www.netflixprize.com/). We have considered both items and users relationships on Netflix dataset for predicting movie ratings. Here, we have followed different ordering strategies for predicting a sequence unknown movie ratings and conducted several experiments.