Movie box office prediction based on ensemble learning

Abstract

The movie box office is now considered a relatively unpredictable short-term experience product. The profits of the film industry are constantly expanding, and more and more investors are engaged in it. But its uncertainty has caused huge losses for many investors. In this paper, film data from 1980 to 2018 were collected on box office mojo, and then, we use machine learning methods, including the Ensemble learning algorithm, to build a predictive model. Results show that the gradient boosting decision tree (GBDT) gives the best performance, of which R2 is higher than 0.995. Experimental results show that the Ensemble learning algorithm is much better than the traditional machine learning algorithm.