Aveek Choudhury, Harshita Ved, Sagar Singh, Sarang Pande
With the emergence of online communities in the past decade, the popularity of social or peer-to-peer(P2P) lending have risen, increasing the accessibility of credit. P2P lending involves the practice of lending money to individuals (or small businesses) via organizations that match anonymous lenders/investors with borrowers bringing new economic capabilities to financing. A key issue in this peer-to-peer loan origination system is balancing credit risk while increasing credit accessibility.
In the course of this project, public data made available by LendingClub, world’s largest P2P organization with over $11B of loan origination, is leveraged to analyze and build an intelligent system to reduce manual effort and time consumed in credit risk analysis process. Once a loan application is filed by a borrower, some key steps in the process include -
● Assigning a proposed interest rate against a loan application by LendingClub
● Categorizing the borrower into specific grades based on his/her profile
● Funding decision made by the investor given the borrower application and relevant features attached by LendingClub
These 3 steps provide an opportunity to integrate data-driven algorithms in order to aid the business process and reduce effort and time. Given this understanding of the business and the dataset, the following approach was decided for the course of analysis -
● Conduct exploratory analysis to draw conclusions based on region-wise loan targets and understand loan term distribution to help LendingClub understand specific markets for expansion and also design schemes with flexible interest rates and longer terms to attract young borrowers.
● Explore regression techniques to identify most suited borrower characteristics driving interest rates at which a loan might be offered for a particular borrower and attempt grade classification to aid in quick loan applicant rating.
● Use of classification techniques to analyze loan characteristics and make accurate predictions of good vs bad loan, thereby establishing LendingClub’s reliability amongst its clients.