A hybrid ensemble machine learning model to predict success of Bollywood movies

Bollywood is a multi-billion industry. Hundreds of films are released every year, where each film is an investment of multi-crores. In terms of awards or marketing it has found a place in almost every country and culture. It also contributes and attracts skilled and passionate people to become entrepreneurs. Therefore, it becomes a need as well as a huge concern of the director, producer and all stakeholders involved in a particular film to know the chances of the success of a film on the box office before its release. To address this concern, a hybrid ensemble machine learning model has been proposed. The model uses data sets collected from various sources, such as Boxofficeindia, cinemalytics, YouTube, etc. The model performed pre-processing on data set, which included handling of missing values with mean, cleaning of data, and removal of text values. Feature engineering has been applied in the model to create a new feature called act_direct to make the model more robust. Further, the effectiveness of the model has been tested in terms of accuracy and the AUC-ROC curve. From the experimental results, it is evident that the proposed model ensures relatively better accuracy compared to some recent state-of-art models.