| As an important indicator of whether a movie can make a profit,boxoffice is affected by the combined effects of many factors and its influence mechanism is more complicated.It is difficult to predict movie box office accurately.There are some shortcomings in the current research on box office predictions,such as relying on social media information,valuate of filmmakers too simply,and failed to explore the value of filmmakers'cooperative relationships.Especially when predict movies'boxoffice before it is released,it is difficult to apply the method of box office prediction based on social media comments and public popularity.This paper implements a movie box office prediction model based on GBRT and social network.First,the factors that affect movie box office,such as movie type,release schedule,director,and actors,are used as input to train the gradient boosting tree model.On this basis,the cooperative relationship between shadow actors is also explored through social network and quantified as node influence as one of the inputs of prediction,and relational data and graphs,which are non-relational data,are combined as features to establish predictions Model and achieve a higher accuracy rate.At the same time,in order to effectively display the movie box office analysis data studied in this article and facilitate users' use,a simple and efficient web movie data analysis system is developed to provide data analysis services.This paper establishes a dataset and conducts experiments on 3000 movies from 2000 to 2019,and 27,367 filmmakers and 5,900 companies.The experimental results show that the model has a good predictive effect and a good performance for domestic movies with a box office of about 1 billion.The average relative error of prediction on the test set is 31.49%.The research work in this article provides a reference for the market value analysis of movies and filmmakers,the method of pre-screening box office prediction and the development of web movie data analysis system. |
PDF Full Text Request