ramos-iyer/Prediction-of-box-office-profit-of-movies-using-hierarchical-clustering-random-forest-ensemble
This research paper aims to predict the box office profit of a movie using a clustering and random forest based ensemble technique. The motivation behind this comes from the need of better predictive techniques in the movie industry owing to its unpredictable nature. The literature survey found meta data and social data to be important factors in predicting a movies box-office success. The use of clustering and random forest ensemble provided an accuracy of 78 percent on the training data and 62 percent on unseen or test data with a correlation of 78 percent between actual and predicted values. The variable importance chart showcased the budget, budget year ratio, release month, run time, genre, collection and review sentiment score to be influential in the profitability of a movie. The business use cases for this model and data considered is mainly for on-demand video service providers and television service companies.