Binary Classification: Peer-to-Peer Loan Delinquency

**Peer-to-Peer lending** marketplace enable the lenders and borrowers to directly deal with each other, facilitating the lending transactions. - In 2014, P2P Platforms in the United States issued approximately $5.5 Billion in loans. - PWC’S analysis indicates the market could reach $150 Billion or higher by 2025 - US peer-to-peer lending platforms’ origination volumes have grown an average of 84% per quarter since 2007. This is a peer-to-peer loan delinquency model that predicts which customers are most likely to default and based on that, marketplace provider can take appropriate measures. It uses binary classification algorithm namely Boosted Decision Trees. It leverages various *AzureML Studio components* as well as *custom R code* for data cleansing and feature engineering **Identifying variables for Classification model** Filter Based Feature Selection, Linear Correlation, Multi Collinearity, Statistical Test, Business Insights **Inputs:** To make loan transactions transparent, Peer-to-Peer Lending marketplace publish timely data. Data include important information on Loan Rejection, Loan Status(Default, Charged off), Economic Status and Demographic. The input data is the LoanStats3a data set from leader of P2P Marketplace Lending Club with total of 42000 rows and 52 variables (features). It is data of borrowers, who availed peer-to-peer loans and include information related to loan i.e *Loan Amount, Term, Grade, Interest Rate, Debt to Income Ratio, Open Account, Total Account*. Other input variables include *home ownership status, purpose,decsription*. **Outputs:** The output of the model is the probability of delinquency. So for a given borrower the model shows the probability that the given borrower will become delinquent or not. **Author:** Saurabh Singh is a Data Science Evangelist @ Cognizant Technology Solutions