Pre-production prediction of movie success: A case study on Iranian motion picture industry

Cinema industry is a gigantic and risky business. Majority of movies fail to recoup their productionbudgets via their income. The need for a decision support system that can give an estimation ofmovie success to investors is vivid. Most of the previous studies have tried to forecast moviessuccess by post-production factors and there are few of them in which the prediction is solelybased on pre-production features. In this regard, to our knowledge there wasn’t any article toscrutinize Iranian motion picture and compare it to the international peers. In this paper, wepredict the success of Iranian movies. To generate these predictions, we employ well-knownclassification models and improve their results by ensemble techniques and use only preproductionfeatures. Studies that have utilized post-production features like word-of-mouthdata, the awards that a movie wins and etc. have reported high accuracies. However, thosepredictions are of little value to the investor whose money has already been spent in theproduction phase. Lack of transparency in Iranian motion picture industry did cause lot ofdifficulties in terms of gathering data. Nevertheless, despite of those difficulties we were able toachieve decent results. It was possible through our novelty in using some feature evaluationmetrics for which it was possible to gather the data.