Predicting Movie Box-Office Success and The Main Determinants of Movie Box Office Sales in Malaysia using Machine Learning Approach

Abstract

A movie's box office is the revenue generated by the movie via ticket sales. Predicting the success of a movie in the box office is never easy. There are many factors that could potentially affect movie box office, such as its reviews and ratings, star power, genre, seasonality, and et cetera. This study aims to explore the most vital factors that influence the Malaysian box office, and to build an accurate predictive model that is tailored to this market using knowledge discovery in databases (KDD) methodology. Movie data will be obtained from FINAS [1], Box Office Mojo [2] and IMDb [3], so that it can be cleaned and processed. The cleaned dataset will be used to build support vector machine (SVM), neural networks (NN) and multilayer perceptron (MLP) models. This paper analyses the efficiency of the three models to predict the box-office success of movies, while analysing the influence of variables. At the end of the study, the most suitable model will be selected. The analysis shows that the most important factors that influence movie box office are movie budget and movie review scores for both local and international movies. In addition, the multilayer perceptron (MLP) model with its accuracy of 0.7529 is the best fit model to predict Malaysia box office for local movies. On the other hand, for predicting the Malaysia box office performance of international movies, neural network (NN) is the best fit model with an accuracy of 0.6171.