Loan Default Prediction: A Complete Revision of LendingClub

Abstract

The study aims to determine a credit default prediction model using data from LendingClub. The model estimates the effect of the influential variables on the prediction process of paid and unpaid loans. We implemented the random forest algorithm to identify the variables with the most significant influence on payment or default, addressing nine predictors related to the borrower's credit and payment background. Results confirm that the model’s performance generates a F1 Macro Score that accomplishes 90% in accuracy for the evaluation sample. Contributions of this study include using the complete dataset of the entire operation of LendingClub available, to obtain transcendental variables for the classification and prediction task, which can be helpful to estimate the default in the person-to-person loan market. We can draw two important conclusions, first we confirm the Random Forest algorithm's capacity to predict binary classification problems based on performance metrics obtained and second, we denote the influence of traditional credit scoring variables on default prediction problems.