Research on Box Office Prediction of Commercial Films Based on Internet Search Index and Multilayer Perceptron

Abstract

The film industry has grown rapidly in recent years. The box office is one of the important indicators to measure the success of a movie. In comparison with traditional film marketing manner, the prediction of box office based on search trends can make traditional marketing more targeted and confident. The innovation of this paper is the search trends, Baidu index and Google trends, is used as an independent variable on various methods of linear regression, moving average method and multilayer perception (MLP) for prediction of box office. The research data is collected from IMDb for north American film market, Endata.com and Maoyan.com for mainland China film market. The performance of box office prediction is compared based on different types of forecasting methods, and with or without the search trends as the independent variable. It is found that the MLP model based on Baidu and Google index has the best prediction performance on the movie box office, with an accuracy of 81.11%. While the prediction error is also evaluated using the mean absolute percentage error (MAPE) calculation, and the result is 18.89% for MLP, 9.53% is improved in accuracy over MLP with search trends. Finally, the search trends on the Internet as an independent variable offer consistent better performance. Hence the film company can pay much attention on Internet marketing to optimize their marketing strategy.