Analysis and Comparison of Loan Default Prediction Models Based on XGBoost and LightGBM Algorithm

Abstract

Based on the default loans caused by information asymmetry and uncontrollable factors at this stage, this paper will use two algorithm models, XGBoost and LightGBM, to extract and screen the relevant information of the applicant and build a loan default prediction model to predict the default situation of the loan. And the two different models were compared and evaluated to provide data reference for financial institutions to select and build the loan default prediction model to reduce their risks and bank losses to a certain extent.