Carlos Estebes

Predicting Startup Outcomes Using Data Science

A startup is a company founded for rapid growth and scale. These types ofcompanies are known to have an extremely high failure rate and wide range ofinvestment sizes and types. In this paper we explored predicting several startup outcomesusing Data Science. We specifically achieved this by utilizing machine learningalgorithms to create models based on numerous startup features. The algorithms exploredincluded k-Nearest Neighbors, Decision Tree, Random Forest, Adaptive Boosting,Gradient Boosting, and Artificial Neural Networks. We were able to achieve actionableprediction results to determine whether a startup will succeed by an exit, fail, or stagnate.These results can help guide investors on making more targeted startup investmentdecisions. We concluded that ensemble learning methods provided the best results onpredicting our startup outcomes.