Research Of Movie Box Office Forecast Model Based On Multi-factor Fusion

The movie box office is an important part of the movie industry,predicting the box office of a movie on the first day of release helps film producers or investors to control risks,helps theaters to formulate film layout plans,and guides the marketing strategies of theaters and production companies.This thesis proposes a new movie box office prediction model to improve the accuracy of movie box office prediction.The main research contents are as follows:(1)Add consumer factors such as movie audiences and the emotional tendencies of movie critics on the first day,and combine movie characteristics and market factors to build a movie box office influencing factor system,then design a reasonable method to calculate the influence index of each box office influencing factor and assign weights to different influencing factors;(2)Connect any two-layer network in the DNN module of x Deep FM by splicing in the feature dimension to improve the combination and interaction of shallow features and deep features,making the model structure more stable and feature utilization higher,and alleviate the gradient vanishing problem.Finally,two sets of comparative experiments are set up to verify the prediction effect of the model proposed in this thesis on the constructed dataset containing 678 movies from 2011 to 2021.The experimental results show that the coefficient of determination(R-Squared)is 0.8403,which is higher than 0.7526 of BP neural network and 0.8166 of x Deep FM;in terms of prediction of well-known movies,the prediction results of the model constructed in this thesis are also closer to the actual value than the two comparative experiments.It can be seen that the research and work of this thesis have achieved good results,and the box office prediction model constructed has improved the accuracy of box office prediction to a certain extent.