Predicting Success of Campaigns on Membership based Patreon Crowdfunding Platform

Abstract

Crowdfunding platforms, such as the Patreon platform, are a means of regular financial support to entrepreneurs and artists who create independent content in the form of images, videos, podcasts, comics, games, or any media that supporters enjoy. Entrepreneurs leverage their potential base of patrons by using various social media platforms. Even though this collaboration has proved to be a practical approach to raising funds, it is difficult to predict the success rates of new projects. In this paper, we consider Patreon as the membership-based platform and our empirical analysis shows that half of proposed projects turn out to be successful. In this research, we build a data analytics approach to predict the rate of success of Patreon projects based on a dataset containing details of various features and historical information about previous projects. We employed a family of supervised classifiers that includes Naïve Bayes, Logistic Regression, Random Forest, and Boosting algorithms to predict the success of a given project. Currently, the Gradient Boosting classifier has achieved an average accuracy of more than 74%. Such results could help creators to define a path to better promote their content and improve monthly pledges.