This project aims to use machine learning techniques to accurately predict the default rate of an individual borrower. In the real business world, a rather accurate prediction will benefit stakeholders since they can better assess early-stage loan performance on historical data, conduct contribution analysis to identify risky segments, and measure credit expansion opportunities.
A detailed version of description and result of the project can be found here: Report
Prosper Market Place Loan Dataset from Kaggle comprising 81 variables and 113,9794 observations from 2005-2014.
Build Logistic Model, ANN, Random Forest, XGBoost for model selection and comparison.
Borrower from Loan with lower risks are long term employer with less delinquencies,inquiries who possess good credit history.