Leveraging historical data from the movie industry, this study built a predictive model for movie success, deviating from past studies by predicting profit (as opposed to revenue) at early stages of production (as opposed to just prior to release) to increase investor certainty. Our work derived several groups of novel features for each movie, based on the cast and collaboration network (who’), content (‘what’), and time of release (‘when’).
Michael T. Lash, Sunyang Fu, Shiyao Wang, Kang Zha