This study investigates the importance predictors for lender’s decisions on peer-to-peer lending based on prospect theory. We apply two machine learning algorithms (decision trees and random forest) to identifying the important measures. We found that borrower’s default history is the most important variable, followed by the default ratio of the borrower’s other type of friend and borrower’s credit score as lender. These results prove that lenders play ‘safety rule’ in the highly risky P2P lending business. Maximum entropy algorithm verifies these results and proves the robustness of our model.
All Science Journal Classification (ASJC) codes
- Business, Management and Accounting(all)
- Computer Science Applications