Neural Network Based Movie Rating Prediction

Abstract

Movie rating is a numerical comprehensive evaluation of movie viewers. It represents an objective basis for general audience and advertisers to choose high quality movie, and is a crucial factor affecting row piece and estimate final box office receipts. As a result, predicting the ratings of a new movie is a great convenience for the audience, cinema, advertisers and investors. This paper identifies the movie rating prediction factors, designs a series of movie evaluation metrics, and proposes a movie rating prediction model based on the neural network algorithm. After preliminary data preprocessing, our algorithm can reach a movie rating prediction accuracy up to 70%. Through the improvement of the starring metric, the prediction accuracy can be further improved to 88.8%. Compared with other related work of movie rating prediction and movie recommendation systems, our method has high accuracy and does not require user information. We compare our solution with several existing methods. Evaluation results show that the prediction accuracy would improve by 89.8%, 48.0%, 50.5% and 37.2% compared with the knn algorithm, the decision tree algorithm, the SVM algorithm, the NBC algorithm respectively.