Mining Trailers Data from YouTube for Predicting Gross Income of Movies

Abstract

YouTube is the most popular video contents sharing platform around the world. As a consequence, YouTube has become one of the most preferred choices to the movie producers and studios for connecting/communicating with their potential viewers through sharing trailers and teasers. Data regarding the trailers of a movie from YouTube can provide useful insights for predicting the gross income of movies. In this paper, we have prepared a dataset of 7988 movie trailers from YouTube. The dataset contains different attributes like opening income, number of views, number of likes, number of dislikes, number of comments. We prepared two prediction models and applied four regression techniques to find out the most suitable technique for predicting the gross income of a movie. The comparative analysis has depicted that linear regression is the most suitable method regarding the prediction of movies gross income using these attributes. Furthermore, we have provided future research issues from where our work has ended.