Loan Default Prediction Based on XGBoost

Select Volume / Issue:

• Volume 11,Issue 1, Jan. 2023

Year:

2023

Type of Publication:

Article

Keywords:

Loan Default Prediction, IV Information, Model Comparison, XGboost

Authors:

Xiaoqian Peng; Jingyi Zhang; Xu Cheng; Yifan Huang; Zhouqiang Zheng

Journal:

IJISM

Volume:

11

Number:

1

Pages:

20-29

Month:

January

ISSN:

2347-9051

Abstract:

Aiming at the credit risk loss of commercial banks caused by loan default, this paper establishes an integrated learning model to predict customer default based on the loan record data of credit platform to reduce credit risk. According to the characteristics of unbalanced loan data categories and high feature dimensions, this paper cleans the data and uses IV information to screen the factors, so as to select the factors with strong predictive ability. Then, on the basis of comparing various models, this paper chooses the best XGboost model to construct the loan default prediction model and evaluate the model. The analysis found that the indicators of XGboost are high, indicating that the performance of the model is good and can be used to accurately predict loan defaults. This provides a reference for commercial banks and other loan platforms to provide credit products to borrowers.

Full text: IJISM_990_FINAL.pdf [Bibtex]