Research On Movie Box Office In China Based On Multiple Regression And Neural Network

With the booming development of the movie industry,the accurate prediction of movie box office has become an effective way of avoiding risks in movie investment and has gradually become the focus of attention in the movie industry and statistics circles.The paper mainly studies the influence factors and prediction of film box office.According to the characteristics of the short life cycle of film,the affecting factors of film and so on,the paper firstly identifies 45 indexes to construct the index system about the influence factors of the film box office.Based on the 1957 films released nationally in the past 2013-2017 years,the influencing factors were analyzed by multiple linear regression,integrated decision tree and integrated neural network.The analysis of the three models shows that the four factors that most influence the movie box office are the duration of movie,the production technology IMAX,the famous publishing organization and the movie rating.Secondly,taking the box office of opening week and influencing factors as input variables,the above three models and recurrent neural network(RNN)model were used to predict the box office of 10 movies in 2017,with a total sample size of 1967.The empirical results show that whether the opening week box office forecast total box office,or the influence factors to predict the total box office,RNN among the four model is optimal(its accuracy is highest).On the other hand,The accuracy of the RNN model,which predicts the total box office by box office of first week,is higher than the accuracy of the RNN model that predicts the total box office with multiple factors.From the above,the best prediction model about the total film box office is based on the RNN model of the first week box office.It has a certain reference value for the prediction of the movie box office in the actual situation.Based on the above results,the important factors of film box office and the movie box office prediction model of RNN are put forward.At the end of this article,some suggestions on the future development of the film industry are put forward.