Study On Film Release Day Box Office Prediction Considering Competition And Consumption Intention

The film's release day box office have an influence on the screening plan of the cinema line,directly affecting the follow-up screening day box office,which has an important impact on the total box office revenue.Without an accurate release day box office prediction,it is difficult for the cinema line to reasonably arrange screening for the release day.A week before the screening was an important time spot for the release day box office prediction.Therefore,it is of great practical significance to propose a release day box office prediction method suitable for the week before the screening.It is difficult to predict the box office on the release day when there are insufficient types of predictive variables and available data.In order to predict the box office on the release day,it is necessary to study the factors affecting the box office and the applicable algorithm.The box office forecast on the release day is a supplement to both box office prediction method and demand forecast method,which has certain theoretical significance.Research on the prediction of release day box office has done as follow.First,find a quantitative approach to film competition and consumption intention before movie release.Unlike the box office forecast after movie release,the actual data of the box office did not appear before the release day,and there was no neither word-of-mouth information.These two types of data are the most commonly used data in the box office forecasting method of the released movie.Therefore,considering quantify competition and consumption intention as prediction variables.The quantify approach is designed in following ways.The film competition is quantified from the high-bay movie and the type of movie in the same period.The potential audience consumption intention is quantified from the explicit and implicit consumption intentions of social network users.Second,a box office prediction method combined Lasso and XGBoost algorithm considering competition and consumption intention was proposed.Combined with the basic information of the film,the movie release day box office prediction variable set was established.Then Introduce a prediction algorithm that combines the Lasso and XGBoost algorithms in box office prediction.Fully consideration of film competition and consumption intentions leads to an increase in variables,which is prone to over-fitting and degradation of prediction accuracy.In the variable selection stage,the Adaptive-Lasso algorithm is used to extract features,reduce the predictor variable dimension,and reduce the training time.The XGBoost algorithm is applied to improve prediction accuracy and prediction stability.Using the collected data of 10 movies,some main conclusions are as follows: using SVM model and CIMM(after domain migration training)to identify consumer intentions in social networks have capable accuracy;Compared with multiple linear regression,support vector regression and other traditional box office prediction methods,the box office prediction method of movie premiere day combining Lasso and XGBoost has an advantage of more than 5% accuracy and lower calculation cost.The prediction accuracy after considering the competition and consumption intention of movies is higher than without considering.The experimental results show that the box office prediction method based on considering competition and consumption intention is more suitable for the box office prediction of film release day.