Early Prediction of Movie Success Using Machine Learning and Evolutionary Computation

The motion picture is one of the major industries in the entertainment domain and a key contributor to the world-wide economy. Millions of dollars are often required and invested in the movie production process. Predicting the rate of success of a movie before its production will avoid huge financial losses. Various approaches exist in the literature to tackle the problem of forecasting movie success. However, most of these approaches fall short in creating an efficient model that could help investors and stakeholders in the decision-making process. These approaches rely on post-production or post-release information, making them inappropriate for pre-investment prediction. Existing approaches that tackle the pre-production forecasting show low predictive performance in general. Due to the white-box nature of decision tree algorithms, practitioners would be interested in leveraging the tree-like structure as a decision-making system while producing a movie. In this work, we propose an evolutionary approach, based on Genetic Algorithms (GA), for optimizing the outputs of the decision tree algorithm (C5) used for the prediction of movie success during the early stage of production. Experiments demonstrate that our hybrid method combining machine learning and evolutionary computation significantly surpasses current state-of-the-art machine learning techniques, achieving a prediction accuracy of 90.5%.