Improvement Of BP_Adaboost Algorithms And Application Research Of Total Box Office Classification Forecast In The First Round Of Financing

The film industry is developing rapidly,meanwhile,most of the films are at a loss and the box office revenue is very different,which poses a great risk to the investment of film investors.If movie investors can predict the movie box office in the first round of financing,it will be very important to ensure the profitability of film investors and the long-term development of the film industry.The film in the first round of financing of the film has fewer influencing factors and cannot accurately predict the movie box office.Therefore,machine learning methods can be used for classification prediction.BP_Adaboost algorithm is a commonly used classification prediction method.BP neural network is used as a weak classifier to compensate for the errors caused by BP neural network subjective selection training samples,and is widely used in medical diagnosis,environmental detection,fault diagnosis and other fields.However,it has not been applied to the movie box office classification forecast.BP neural network has the disadvantages of low learning efficiency,slow convergence rate and easy to fall into local optimum.BP neural network is also a weak classifier of BP_Adaboost algorithm,and the above problems also exist.In view of the above shortcomings,this paper proposes three algorithms to improve the BP_Adaboost algorithm and apply it to the total box office classification prediction in the first round of financing.The following work was specifically carried out.Firstly,the applicability of the total box office classification forecasting based on the BP_Adaboost algorithm for the first round of film financing is studied.Targeted research on the factors affecting the total box office in the first round of financing of the film,and selected 11 variables under the four influencing factors,including film producer,master producer,issuer and script,as the input variables,and classify the total box office of the movie into four Levels,with the classification value as the output variable.The total box office classification prediction method for the first round of financing based on BP_Adaboost algorithm is constructed.The sample data set is used to verify the applicability of the method.Then,three algorithms are proposed to improve the BP_Adaboost algorithm and apply it to the total box office classification prediction in the first round of financing.Mind Evolutionary Algorithm(MEA)and Levenberg-Marquardt method(LM)can improve the BP neural network.Therefore,three methods such as MEA,LM algorithm and MEA-LM algorithm are considered to improve BP_Adaboost algorithm.And use the film sample dataset to verify the effectiveness of the improved algorithm.Finally,comparing the overall performance of the improved algorithm with other algorithms,BP,BP_Adaboost,MEA-BP,LM-BP,MEA-LM-BP,MEA-BP_Adaboost,LM-BP_Adaboost and MEA-LM-BP_Adaboost are used in compareing in respect of model accuracy,model stability and model K-fold cross-validation.The results show that the BP_Adaboost algorithm can be used to predict the total box office classification of the movie in the first round of financing,and the prediction accuracy of the BP_Adaboost algorithm improved by MEA,LM algorithm,combined with MEA and LM algorithm is improved,which proves that all three improved algorithms are effective.The MEA-LM-BP_Adaboost algorithm has a total box office classification accuracy of 73.3%.The overall performance of 8 prediction models is compared.The comparison results show that the proposed BP_Adaboost algorithm combined with MEA and LM algorithm has the best overall performance.In addition,the LM algorithm has a good effect on improving the accuracy of model prediction,and the MEA has the effect of improving the stability of the model.