Explainable AI for Credit Scoring: Balancing Model Interpretability and Predictive Power in Fintech

Views: 3
Downloads: 0
DOI https://doi.org/10.55923/aksara.v5i1.4
Tari Kurniawan 1 , Doni Pratama 2 , Nur Lubis 3 , Anisa Nasution 4
1, 2, 3, 4Faculty of Information Technology, Universitas Diponegoro
1 tari.kurniawan@undip.ac.id , 2 doni.pratama@undip.ac.id , 3 nur.lubis@undip.ac.id , 4 anisa.nasution@undip.ac.id

Downloads

Black-box machine learning algorithms in credit scoring violate regulatory mandates for adverse action notifications. Combining gradient-boosted decision trees with TreeSHAP feature attributions, we establish an auditable scoring engine that meets compliance demands without sacrificing area under ROC curve.
Issue Vol. 5 No. 1 (2026)
Section Articles
Pages 308-321
Published 2026-03-15

Most read articles by the same author(s)

No other articles found by this author.