Segmentation of Problematic Loan Customers Using The K-Means Clustering Algorithm to Support Strategic Decision-Making (Case Study: Bank Mega Finance Bengkulu)

Novrian, Willi
Afriani, Annisa
Sari, Julia Purnama
Putra, Yusran Panca

Indonesian Journal of Computer Science and Engineering
Vol. 2 No. 01 (2025): IJCSE Volume 02 Number 02, November 2025

This study analyzes 5,305 records of non-performing loan customers from Bank Mega Finance Bengkulu using the K-Means Clustering algorithm within the CRISP-DM framework. Based on variables such as tenure, outstanding balance, installment amount, and payment delay duration, the analysis identified three customer risk clusters (high, medium, and low) with a Davies-Bouldin Index (DBI) of 0.201, indicating good clustering quality. The segmentation results can help determine collection priorities, loan restructuring, and risk mitigation strategies, demonstrating the effectiveness of data mining in supporting strategic decision-making in banking risk management.