Objective
Our goal is to create value by improving the efficiency of credit risk management and offering users better customer experiences, as well as higher returns. We are implementing innovative machine learning techniques that have not been extensively explored in the context of risk classification.
Our team really gained an appreciation of harnessing the differing strengths of each member in order to achieve an overall goal. Setting and holding each other accountable for specifically defined deadlines helped to drive the success of our project. Additionally, we learned the value of documenting our individual work in a sufficiently detailed way in order to make it collaboratively useful to others. The machine learning techniques that and the optimization process of those techniques have an abundance of real-world applications, which will undoubtedly serve us well as we begin our careers after graduation.
— Becca Menke
Project Description
Prosper is a peer-to-peer lending platform that connects borrowers directly to lenders, offering more attractive rates for both parties. In order for companies like Prosper to facilitate loans with such low rates, it is imperative to be able to accurately assess the risk associated with each potential loan. A good classifier identifies dangerous borrowers, protecting investors from large losses, while at the same time recognizing good borrowers. Our task is to use machine-learning techniques and other advanced statistical methods to develop a new risk classifier minimizes the number of customers that are misclassified.
Accomplishments
When we started working on the project, none of us had a very strong background in machine learning. As a part of the literature review for this project, we read through books and took online courses in machine learning and related topics to increase our understanding. We have trained a variety of models using large amounts of data and evaluated the success of each one in comparison to Prosper’s existing model. Some of our models that use standard classification techniques, are even outperforming Prosper’s current model.
Apart from gaining technical expertise, we have learned a lot about team dynamics and efficient collaboration through our work on this project. We have experienced first-hand the value of being accountable for individual work, as it contributes to an overall goal and the pride in celebrating successes thereof.
Open communication and sharing knowledge are essential when we work in team and need to deal with many technical difficulties. The questions one team member has might also be other teammates’ concerns, and it is really helpful and much more efficient when we as a team discuss the issues together than trying to solve problems by oneself. Collaboration leads to innovation and creativity. Last but not least, I really appreciate the help of my advisors and teammates, who supported me all the way here.
— Sonia Yao
Lee Fleming – Faculty Advisor
The Prosper Team provides an excellent example of a perfect capstone. First, we have a motivated and professional team that has worked very hard and collaborated in a very effective way. Second, we have a fantastic industry advisor in Brad Pennington, who has outlined the social and economic challenge and at the same time, guided the team to apply their engineering expertise to solve that social and economic problem. Finally, the capstone is part of a larger effort between the Fung Institute and Prosper and the CrowdFunding community; this effort includes support of research on CrowdFunding and Entrepreneurship from the Kauffman Foundation.
Brad Pennington – Industry Partner
VP Risk Strategy and Analysis, Prosper Marketplace
Working with our capstone team and Lee Fleming has been a great experience. The students engaged the problem with open minds, a great attitude and a desire to apply their classroom learning to a real-world challenge. The entire project was excellently supported by the Fung Institute’s professional and highly-engaged staff. An important part of building an ecosystem that brings borrowers and lenders together is to be personally invested in pushing the technological boundaries that divide them. Prospers’ involvement with capstone is proof of our commitment to pushing those boundaries.
← View all Capstone Projects