COMPARATIVE ANALYSIS OF MACHINE LEARNING MODELS FOR PREDICTION OF MOVIE’S BOX OFFICE SUCCESS

초록

The film industry is a high-risk, high-return business from an economic point of view. It is necessary to lower the uncertainty of the film industry by predicting the success or failure of the film business before the film’s release. This study builds the various models to predict the success of films before the release in order to give helpful information to the film industry in the production stage. After selecting variables, such as director, actor, genre, runtime, grade, language, cost, original story, release month, promotion, and marketing before the movie is released, we visualize and analyze the variables. We implement six predictive models using various algorithms to transform the data to fit the predictive model by performing a pre-data processing process. We compare and evaluate the performances of the predictive models on a variety of criteria, such as predictive accuracy, precision, recall, f1-score, and learning time. We visualize the performance of the models. The deep neural network model shows the highest prediction accuracy, and the random forest model also has relatively high accuracy.