Analysis And Prediction Of Influencing Factors Of Movie Box Office Based On Fully Connected Neural Network

In recent years,with the country' s strong support for the cultural industry,the scale of the film market has gradually expanded,and movies have become a necessity for our modern lives.At the same time,with the rise of social media,the number of information resources has exploded.How to find useful knowledge in the massive information has become an urgent problem for the majority of researchers.Movie box office not only reflects the national economic level,but also reflects the audience' s value orientation and the attitude orientation of internet public opinion,which has an impact on movie box office.Analysis and forecasting can not only reduce the investment risk of the film market,but also attract the attention of investors.Therefore,box office forecasting research is a challenging but very important task in the decision-making process of the film industry.Based on the previous research results,this article processes and analyzes the Movie Database(TMDb)data set provided by the Kaggle competition.The data set includes information about nearly 5,000 movies in western countries from 1916 to2017.The main work contents are as follows: First,Python software is used to clean,process and screen the data,so as to retain the complete data with research significance.Then it makes a visual analysis of the influencing factors related to the box office,and analyzes the characteristics that may affect the box office through word cloud map,Top10 box office,Top10 score,and frequency distribution chart of film type.It focuses on the different influences of film type,director,leading actor and schedule on the box office revenue.Then machine learning algorithm was used to establish a Fully Connected Neural Network(FNN)prediction model on the data set,in order to show the accuracy and superiority of this model in the prediction task in this paper,Root Mean Square Error(RMSE)was taken as the evaluation index.Compared with the prediction results of XGBoost(e Xtreme Gradient Boosting)model and Lightg BM(Light Gradient Boosting)model,the prediction effect of the fully-connected neural network model is finally verified to be better,so it can be considered to be applied to future box office prediction.The significance of this move lies in that it can apply the development experience of western film box office to the domestic film market,thereby reducing the investment risk of the domestic film market and providing important support for the business intelligence decision-making process such as the production,distribution,screening of film projects and the development of related products.Based on the research results of this paper and combined with the actual situation in China,the following suggestions are put forward for the development of China's film market: adjust the proportion of film types appropriately,and increase the types of films that audiences like(such as comedy,action film,etc.);before the film shooting and production,according to different content to choose the appropriate director and actors;reasonable arrangement of screening time and so on.