Improved Peer-to-Peer Lending Credit Scoring Mechanism using Machine Learning Techniques

Abstract

Peer-to-Peer(P2P) financing is a fast developing modern financial exchange network, which bypasses conventional intermediaries by linking lenders and borrowers directly. However, the online P2P lending platforms are faced with a problem of information asymmetry between lenders and borrowers. Assessing borrower’s creditworthiness is important because many P2P loans are not secured by collateral. Banks use credit scoring to evaluate borrower’s creditworthiness and reduce potential loan default risk. However, in P2P lending platform effective credit scoring models are hard to build due to insufficient credit information. This work is based on an empirical study by using the public dataset from the LendingClub, one of the largest online P2P lending platform in the USA. The aim of this study is to investigate the influential factors on loan performance on the basis of the credit score in the online P2P lending industry. This work improves the online credit scoring models and gives insight into the specific determinants that are influential for the score