Prediction Research Of Film Box Office Based On BP Neural Network

With the continuous development of science and technology,the combination of film and modern technology,more and more diversified films have entered the film market.Although film is an art form of spiritual culture,it is ultimately its commercial value.There are too many complicated factors affecting the box office,but the box office directly affects the development of the Chinese film market,so the prediction of the Chinese film box office is particularly important.It is difficult to predict the box office of a movie.The reason is that the factors are too complex and the prediction method is not good.If the box office data of the upcoming films can be predicted more accurately,it can effectively help the film producers to make more reasonable decisions and timely adjust the film distribution.This is the starting point of this paper.In this paper,according to the basic principles of constructing index system,the influence factors of film can be divided into the film itself characteristics influence factors and the factors influencing the user behavior,the 15 indexes are selected,and the influence factors of the selected indicators can be divided into three categories,respectively is a type of index,multitype,influence,to quantify the different types of indicators,respectively.Then through art data platform,such as selecting the Chinese market in the 490 Chinese film more than fifty million film empirical prediction analysis,rough set attribute reduction by removing relatively unimportant index,fuzzy classification,clustering,respectively,set up the BP neural network regression prediction model,is obtained by 5 fold cross validation method to optimize prediction results,and each index is obtained by reducing index respectively the influence degree of the prediction precision of the model,and then compared with the BP neural network and support vector machine(SVM)model prediction effect.Finally,the new film "only nonsuccess know" box office forecast,according to the results of the rationalization proposals,and analysis of some publicity indicators on the impact of the film box office.The following conclusions are drawn:(1)Compared with the normal BP neural network model and support vector machine model,the BP neural network model processed by rough set attribute reduction and fuzzy clustering is more accurate in predicting the box office of films.(2)Format,region,award or nomination and producer influence have a small impact on the box office of a film.Audiences pay less attention to these aspects of the film,so the film producers and distributors can moderately reduce the time and energy spent on these aspects.(3)The influence of the leading actor has a great impact on the accuracy of the model,which is much higher than other indicators.The first-week box office and Baidu index,which are publicity and distribution properties,also have a great impact.The sequel and popular IP in the story familiarity are the ones with the least impact.(4)Baidu index in the properties of movie promotion and distribution can improve the box office more than the box office of the first week and Douban score.Through the conclusion and the box office prediction model,it has certain guiding significance for the film producer to use the limited resources to produce the maximum box office revenue,it can also help the film distributor to make the publicity plan,and it can provide certain reference for the theatrical release.