Applying Machine Learning Techniques To Maximize The Performance of Loan Default Prediction

Abstract

In peer-to-peer (P2P) lending, borrowers would access loans with lower interest rates than what they usually got from traditional lenders. People can directly borrow from the P2P platform with the rules that make them easy to borrow loans and invest free funds into P2P, which can benefit both borrowers and lenders. However, the easy way to borrow loans comes with risks. One of the major issues is that borrowers may default on the loan taken. In such cases, they can get loans quickly from P2P online platforms without any bank interferences. Thus, the lender can calculate his risk for loan default. In this project, we consider the P2P lending data to predict the loan default reassuring the lender to continue providing loans in the future. In our analysis, we consider the Logistic Regression, Naive Bayes, Random Forest, K Nearest Neighbour, and Decision tree to classify loan data based on their likelihood of default. The simulation result in our algorithm provides a significant accuracy of 94.6%.