Research On Movie Scoring Prediction Model Based On Machine Learning

Movie scoring is an important index to evaluate the quality of film production and audience reputation.It is affected by the internal factors and external environmental factors of the film itself,and has a more complex influence mechanism.It is difficult to predict the movie rating,especially before the film is released.After the glory of the 1990s and the low tide of the early 21st century,the mainstream film gradually began to have an important impact on the film market with a number of high-score films that have won a good reputation among the audience in recent years.The deficiencies in the prediction research of film score mainly include the great dependence on the evaluation of viewers and the popularity of public opinion,the single quantitative method of the internal factors of the film,the lack of special research on the mainstream film.The existing prediction based on external evaluation data is difficult to use before the film is released.This paper collects the mainstream films that have been publicly released since the 1990s,and quantifies the variable factors affecting the score,such as director level,screenwriter ability,main actor performance and film type.According to the characteristics of the mainstream film,two new variables are added to form the training data set and use it to train three models:decision tree,random forest and support vector machine.Through the analysis of the prediction results of the three models,it is proved that the random forest model is more suitable for film score prediction in the above three models.The random forest model not only has the best effect on predicting the score of theme movies,but also can obtain the main factors affecting the score of theme movies by analyzing the model.In order to further verify the effectiveness and generalization of the model,this paper collects the relevant information of 15 mainstream film released in the last three months,uses the rating prediction model to predict,obtains 80%of the overall accuracy,and has a good prediction accuracy for mainstream film with a rating of more than 7,indicating that the model in this paper has high performance.On the one hand,the research work of this paper provides the audience with a reference to the film quality before watching the film.On the other hand,it also provides a certain reference for the producers of the main melody film in the selection of directors,screenwriters and actors.