Predicting Movie Market Revenue Using Social Media Data

The amount of data being created and processed daily has grown exponentially with the introduction of the internet and social media. While the data are available, there is a struggle to determine how to effectively use and interpret the data. One of the most popular uses for the large quantities of data is to create models to predict the behavior or tendencies. One important application of prediction is predicting financial outcomes using datasets. As a specific case, this study focuses on the use of Twitter data collected leading up to a movie's opening weekend to predict its revenue over the course of each of the opening weekends. Due to the lack of readily available data, the data must be first gathered weekly using Twitter's API and related third party libraries. Construction of the predictive model is based on several machine learning algorithms using a set of features derived from user tweets. The results show that our predictive model can be used to determine the success of movies during the opening weekend by prediction the gross per day value. The modelling and process are presented such that they can be used as an aid to create similar models using other popular social media networks and their data.