Predicting attributes based movie success through ensemble machine learning

Abstract

The film industry has grown into a multi-billionaire industry in terms of entertainment. The success of the film industry depends on the criteria that how much profit a movie would make which gives the tag of a ‘hit’ or a ‘flop’. Predicting the success is guided by various factors like genre, date of release, actors, net gross and many more. Understanding the stakes involved with a movie release that can affect its success or a failure, before-hand can be a great step towards the expansion of the film industry business. Therefore, this study proposes an ensemble learning strategy as a solution to analyze such understanding where predictions from previously guided attribute calculations can be used to enhance future success/failure accuracy. This study shows various strategies used in the literature to analyze and compare the results obtained. The various machines learning algorithms SVM, KNN, Naive Bayes, Boosting Ensemble Technique, Stacking Ensemble Technique, Voting Ensemble Technique, and MLP Neural Network are applied on the dataset to predict the box office success of a movie. The paper uses various algorithms and their trends in predicting the outcome of a movie and shows that the proposed methodology outperforms the existing studies. The most effective algorithm in the study is Gradient Boosting with a success rate of 84.1297%.