Development and Evaluation of Machine Learning Model for Loan Default Risk Assessment

View Full Paper

Ask AI

Bookmark

Authors

Discussion

This research demonstrates improved loan default prediction accuracy in a real-world dataset, highlighting effective data techniques.

The Random Forest model achieved a test accuracy of 96.26%, providing reliable predictions on loan default risk.
Using Synthetic Minority Oversampling Technique improved model sensitivity to defaulters in an imbalanced dataset.
Evaluations were conducted using stratified 5-fold cross-validation, focusing on metrics sensitive to class imbalance.
Recommendations aim to promote the adoption of interpretable predictive systems in financial institutions.

UKPABI et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark