Loan Risk Predictive Modeling on Lending Club Data

Elena Tumanov

Master's thesis

Abstract:

In the lending industry, the loans are provided by investors to borrowers in exchange for the repayment with interest. In case the borrower repays the loan, the lender would have a profit from the interest. However, in case the borrower fails to repay the loan, the lender loses money. Predicting the risk if the borrower can be unable to repay a loan is the most common problem faced by the lenders. There is an immense risk while making a decision on loan approval. Therefore, the goal of this diploma thesis is to gather data about loans from the Lending Club which is a peer-to-peer lending platform based in USA, and use machine learning techniques in order to extract meaningful information and predict if a borrower would be able or not to repay the loan, if the borrower would be Charged-off or not. In this diploma thesis, several Machine Learning models such as Random Forest, Logistic Regression and K Nearest Neighbours are trained to determine if the borrower has the ability to repay its loan. Furthermore, the performance of the models was also analyzed. As a result, the Logistic Regression model is chosen as the most optimal predictive model and it is likely that the Interest Rate and the Term significantly influence the forecast.

Keywords

Predictive Modelling, Exploratory Analysis, Loan Risk, Peer-to-Peer Lending, Linear Regression, Big data, K Nearest Neighbors, Random Forest

Jazyk práce: English

Datum vytvoření / odevzdání či podání práce: 28. 11. 2019

Obhajoba závěrečné práce

• Vedúci: Ing. Tomáš Hlavsa, Ph.D.

Czech University of Life Sciences Prague

Faculty of Economics and Management

Master programme / odbor:
Systems Engineering and Informatics / Informatics