<![CDATA[This study aims to segment customers based on 2024 m-banking transaction data at a regional bank in North Sumatra using the K-Means Clustering algorithm. The research process follows the CRISP-DM stages, including data preparation, data mapping, data cleaning, and clustering using Python. From 1,035,184 transaction data, 1,024,767 valid data were obtained, grouped into several service categories (e-wallet, internet, electricity & water, and insurance). The clustering results show different customer behavior patterns in each category, ranging from micro customers with small transactions and high frequency, regular customers with medium transactions, to premium customers with large transaction values but low frequency. The results of this segmentation can help banks design more targeted marketing strategies, improve operational efficiency, and support data-based fraud detection systems.]]>
Copyrights © 2026
