Research on loan default prediction based on logistic regression, randomforest, xgboost and adaboost

Guangdong University of Technology, Guangzhou, 510006, China

Corresponding author: 3121004707@mail2.gdut.edu.cn

Abstract

Lenders often experience loan defaults, resulting in huge losses to lenders. Lenders are required to conduct a credit assessment of borrowers before making loans. Machine learning plays an essential role in loan credit analysis. This study analyzes the application of machine learning in loan credit analysis through a dataset of borrowers from Kaggle and looks for an excellent algorithm.This study use Logistic Regression, randomforest, XGBoost and AdaBoost to fit the dateset and compare their accuracy in prediction.In terms of results, XGBoost performed well while logistic regression performed poorly. For banks or lending institutions, using Gradient Boosting Decision Tree like XGBoost to predict loan default can increase profit.