Thesis or dissertation
Date of this Version
Equity crowdfunding is an increasingly popular means of raising capital for early stage startups. It enables entrepreneurs to finance their companies with smaller contributions from a variety of people. This paper studies the relationship between the characteristics of a given company and its ability to raise funds on an equity crowdfunding platform. A series of statistical and machine learning models are fit to data from a U.S.-based equity crowdfunding website, including a logistic regression, a CART decision tree, a naïve Bayes classifier, and a support vector machine. This study demonstrates that a connection exists between the probability of a company’s crowdfunding success and its previous funding history, Twitter presence, media buzz, size, location, and its founders’ educational backgrounds. As a whole, however, the classification quality of the various models leaves something to be desired. This suggests the need for additional data inputs and more longitudinal research in the field of equity crowdfunding.
venture capital, equity crowdfunding, machine learning
Beckwith, J. (2016). "Predicting Success in Equity Crowdfunding," Joseph Wharton Scholars. Available at http://repository.upenn.edu/joseph_wharton_scholars/25
Date Posted: 10 August 2016