Predicting Movie Box-office Based On Deep Learning Of Convolutional Neural Network

With the rapid development of national economy,people's spiritual and cultural life is enriched.Film is an important part of cultural life,it not only enriched people's amateur life and spiritual world,but also to promote the development of China's film and televis io n industry.Movie is a mainstay of China's cultural industry.It brought huge economic benefits to the development of the country,and film box-office is the most important and visualized index to measure the film economic benefits.The development of Chinese film market rapidly,film grossed 45.712 billion yuan in 2016.Therefore,It has important reference value to the investment decisions of Chinese movies that study the influence factors of Chinese film according to the characteristics of the Chinese film market.There are widely literature of movie box office forecasting research,but most of them are in the traditional statistical method and the traditional neural network level.Multiple linear regression can well estimate the influence degree of each characteristics of the film.Also,it can estimate the movie box office and it has good interpretability.But it has a disadvantage which is independent among characteristics.Neural network model for the forecast of film is more accurate,error smal er,but the characteristics to explain the influence degree of the film is not good.The aim of this paper is based on the Chinese film market features,according to the historical movie box office data,I try to put forward the convolution neural network of deep learning model which is applied in the movie box office forecasting.Thesis mainly use film sample data to create data model using multiple linear regression and BP neural network and convolutional neural network(CNN),and study the various influence factors on the film box office,and Compare the advantages and disadvantages of each model.The experimental results show that BP neural network and convolutional neural network prediction is better than multip le regression,but deep learning can get more learning feature between variables,which can more effectively improve the prediction accuracy.