| In recent years,China’s cultural industry has developed rapidly,especially the film industry ushered in spring,and the box office revenue has increased year by year.However,with the rapid increase of profits,competition among different companies in the film industry is also increasing.Research and prediction of film box office is beneficial to film companies in the fierce competition to seize the first opportunity to achieve better and faster development.At the same time,it can also help optimize the publicity and distribution operation,guide cinemas to arrange films,hedge investment risks,and enhance the box office competitiveness of Chinese films.This paper mainly studies and analyzes the prediction problem of film box office.Taking the top 50 films of 2020 box office as an example,under the premise of the previous box office prediction index system,the main theme and epidemic influencing factors are introduced to establish a new box office prediction system.The new box office prediction system selects 31 influencing factors under 8 variables(movie type,influence of leading actors,production country or region,whether it is a major drama,number of confirmed epidemic cases,schedule,distribution company and audience expectation).In this film box office prediction system,firstly,the film box office prediction is studied based on convolutional neural network.Then,combining the advantages of convolutional neural network and support vector machine,a multi-scale convolution based support vector machine(MSC-SVM)model was proposed and applied to film box office prediction.The influence of different parameters on the model performance was analyzed to improve the prediction performance of the model.Through empirical analysis,the following conclusions can be drawn:(1)in this paper,convolutional neural network is used to predict movie box office,and it is proved that the prediction performance of double hidden layer convolutional neural network is better than that of single hidden layer convolutional neural network,and the number of convolutional kernels has an impact on the prediction performance of convolutional neural network;(2)This paper proposes a MSC-SVM model that can be used for film box office prediction.This model combines the feature extraction ability of multi-scale convolutional network and the discriminant ability of support vector machine model,and achieves better prediction results than convolutional neural network and support vector machine.(3)It is feasible and practical to include main melody characteristics and epidemic factors into the box office impact index system.In this paper,the film box office analysis and prediction research,the establishment of a new film box office impact index system,put forward a more appropriate prediction model and complete empirical analysis,can provide guidance for the early release and publicity of the film,has a certain theoretical and practical significance. |
