An Empirical Study On Movie Box Office Forecasting Model

As the mainstay of the cultural industry,the film industry has brought great economic and social benefits.And the box office is the most important indicator measuring film economic benefits.Chinese film is entering the stage of rapid development.Up to now,the total movie box office in 2016 has been close to 50 billion,and China has become the world's second largest movie box office market after the North American.Because the film has the feature of high input and high risk,study of the influence factor and the influence degree of movie box office is means to ensure the revenue of movie distribution and control the risk of movie distribution,and it is of great reference value to make film investment decisions.The researches on box office in China are relatively late,and most of them are theoretical researches.There are few empirical studies.Most of the related researches have different measurement standards and different primary focus,and some of the researches have too few sample data or exist time limit,so some conclusions are not objective.On the basis of the domestic and foreign scholars' researches,this paper chosen a number of reasonably feasible indicators to construct the index system of movie box office influencing factors,which included both the factors based on Internet search and the traditional factors.By using the 631 domestic movies between 2014 and 2017 as research sample data,statistical analysis was carried out and a number of data mining models based on the machine learning algorithm,such as regression,neural network,support vector machine and decision tree were established,as well as model evaluation and empirical analysis.The research has a strong timeliness and pertinence.Finally the decision tree C5.0 was proved to be the best predictive model,and combined with the model result the movie box office forecast factors were analyzed.Through the study I put forward my own reasonable suggestions for the film industry in China,and provided reference value for film investment decision-making.