Predicting the Default Borrowers in P2P Platform Using Machine Learning Models

The online P2P platform's major advantage is that people can borrow or lend money free of intermediary interference. Prediction of the credit risk by the platform should ensure the borrowed money's repayment. This research used Random Forest (RF) in comparison with other machine learning (ML) techniques like Logistic Regression, K-Nearest Neighbor, and Multi-Layer Perception to predict the default borrowers. Lending Club's dataset is utilized for training and analyzing ML models. Statistical measures such as accuracy, recall, precision, F1-score, and the ROC curve are used to compare the data obtained in this study. The results were in accordance with Logistic Regression with the highest precision of 0.95 and RF with the highest AUC of 0.94. This study provides an overall understanding of different models and their prediction of default borrowers. Comparison of these models helps us to identify the most suitable model for the P2P platform.