EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Explainable Artificial Intelligence for Loan Default Prediction and Credit Approval Decision Support

Changarampatt Manoj

Computer and Decision Making: An International Journal · 08.11.2026 · 3 (8S) · 236-244

DOI: 10.1922/comdem.v3i8s.19169 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computer and Decision Making: An International Journal, online erschienen am 08.11.2026. Verfasst von Changarampatt Manoj. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This study aims to design and develop an XAI model for loan default prediction and credit approval decision support using a credit-risk dataset of 32,581 borrower records and 12 variables. After removing duplicate and implausible observations, 32,409 records were analyzed. Age, income, home ownership, employment length, loan purpose, loan grade, loan amount, interest rate, loan-to-income ratio, previous default history, and credit-history length were predictors, with the binary outcome being loan default. The models used were Logistic Regression, Decision Tree, Random Forest, XGBoost and Support Vector Machine, all of which were trained via an 80:20 stratified split and five-fold cross-validation. The model accuracy, precision, recall, specificity, F1-score, ROC-AUC and confusion matrices were used to evaluate the model performance. XGBoost had the best performance in terms of discrimination with an ROC-AUC of 0.948, recall of 81.45%, and F1 of 80.21%. Annual income, interest rate and loan-to-income ratio were the biggest factors affecting default according to SHAP analysis. Among borrowers, risk stratification also subdivided them into low-risk borrowers (3.12% observed default rate), moderate-risk borrowers (12.58% observed default rate), and high-risk borrowers (70.57% observed default rate). The results show that explainable machine learning offers accurate and interpretable support for transparent credit risk assessment without replacing the final decision-making by a human. The framework thus provides a balanced solution, with the ability to explain the results to the borrower, while also supporting risk-informed decision-making for the lending institution.

Zitieren

APA

Manoj, C. (2026). Explainable Artificial Intelligence for Loan Default Prediction and Credit Approval Decision Support. Computer and Decision Making: An International Journal, 3(8S), 236-244. https://doi.org/10.1922/comdem.v3i8s.19169

Vancouver

Manoj C. Explainable Artificial Intelligence for Loan Default Prediction and Credit Approval Decision Support. Computer and Decision Making: An International Journal. 2026;3(8S):236-244. doi:10.1922/comdem.v3i8s.19169

BibTeX
@article{Manoj2026,
  author = {Manoj, Changarampatt},
  title = {Explainable Artificial Intelligence for Loan Default Prediction and Credit Approval Decision Support},
  journal = {Computer and Decision Making: An International Journal},
  year = {2026},
  volume = {3},
  number = {8S},
  pages = {236--244},
  doi = {10.1922/comdem.v3i8s.19169}
}
Kostenlos starten →