An Ensemble Deep Learning Approach to Explore the Impact of Enticement, Engagement and Experience in Reward Based Crowdfunding

Abstract

The exponential growth and ubiquity of the Internet since its advent have led emerging entrepreneurs to seek new avenues of financial support for funding their innovations. Evolved as an alternative source of finance for growing businesses, Crowdfunding democratises venture capitalism by providing benefits to patrons by offering incentives. Although there exist apparent benefits for a funder if a deal succeeds, there are many factors that can contribute to failure. This paper identifies three key factors: Enticement, Experience and Engagement; extrapolating hidden metrics from these factors by analysing Kickstarter projects to explore the effectiveness of content and its sentiments, rewards and their tangibility, funder belief and founder-funder engagement in its success. To validate the discovered metrics, an ensemble deep-learning model that combines textual and numeric features is proposed to predict campaigns with the greatest potential to achieve funding with 93\% accuracy.

Keywords: Crowdfunding, Reward Based, Decision Making, Topic Modelling, Ensemble Deep Learning, Success Prediction