Research On The Hybrid Prediction Model Of Online Ratings Of The Films In Douban Website

With the rapid development of society and the improvement of people’s living standard,people’s demand for spiritual life is rising.As a fast moving consumer goods,the film provides a rich and colorful spiritual enjoyment for people’s leisure life.The online score of the film can provide reference and basis for the theater,the audience,the surrounding industries,etc..so the prediction of the film online score has always been a concern of the film industry.Therefore,how to accurately predict the score of a movie online,for the development of the theater and the surrounding industry is essential,which is also crucial for the audience.Because the online score is difficult to forecast,the empirical research on the prediction of film online rating is also very scarce,and has not formed a set of mature and scientific prediction method.In order to further improve the prediction accuracy of the online rating by using the feature information of the film,a new hybrid prediction model based on the feature information of the film is proposed.In this paper,a hybrid prediction model is designed,and is combined the users’ ratings and the similar film recommendation information from douban website.The model is an effective way to improve the traditional prediction model.At the same time,in order to select the machine learning method and the film characteristics suitable for the film online rating prediction model,to further improve the prediction accuracy of the model,this paper selects more than ten kinds of machine learning models,and makes some comparative experiments.In order to evaluate the hybrid prediction model of the film online rating,this paper selects the film data from the douban website,and uses RMSE index to test.Firstly,this paper introduces the research background and significance of the film rating prediction,and analyzes the current situation of the research on the common data sources and the film prediction model.In the following chapters,this paper introduces the relevant theories of the data sources,the factors affecting the film ratings,and the commonly used machine learning models.Then,this paper puts forward a hybrid prediction model of the film rating on the douban website,and analyzes the hybrid prediction model.Finally,the main work of this paper is summarized,and the shortcomings are put forward.