Using machine learning forecasts movie revenue

Abstract

A successful movie is determined by many factors, and the office box revenue of a movie not only represents its audiences' recognition but also brings social impacts and commercial boons. Traditionally, movie investors need to consider the risks and benefits when deciding whether to invest in movies. In this case, an accurate and reasonable prediction of a movie can help investors reduce the investment risks to a large extent. Therefore, this project focus on using machine learning to build movie revenue prediction models and using basic theories to prove the validity of each model and compare their performance. Besides, by analyzing the importance of each feature, this project can also give some practical suggestions to the film producers to adjust the strategy reasonably in the process of film shooting, production, publicity, and release.