https://www.frontlinejournals.org/journals/index.php/gs-indexing/issue/feed 2025-03-06T08:15:08+00:00 American Research Index Library index@frontlinejournals.org Open Journal Systems <p>American Research Index Library</p> https://www.frontlinejournals.org/journals/index.php/gs-indexing/article/view/701 2025-03-06T08:15:08+00:00 Md Amran Hossen Pabel amran@scientiamreearch.org Biswanath Bhattacharjee bhattacharjee@scientiamreearch.org Sonjoy Kumar Dey sonjoy@scientiamreearch.org Sakib Salam Jamee jamee@scientiamreearch.org Md Omar Obaid obaid@scientiamreearch.org Md Sakib Mia sakib@scientiamreearch.org Sajidul Islam Khan sajidul@scientiamreearch.org Mohammad Kawsur Sharif mohammad@scientiamreearch.org

<p>This study evaluates three machine learning clustering algorithms—K-Means, DBSCAN, and Hierarchical Clustering—for customer segmentation in the banking sector. Using a dataset of customer demographic, financial, and transactional data, we compare the algorithms based on the Silhouette score and Davies-Bouldin index. Hierarchical Clustering performed best, achieving the highest Silhouette score (0.68) and the lowest Davies-Bouldin index (1.15), indicating well-defined and compact clusters. K-Means showed reliable performance with a Silhouette score of 0.62 but required predefined clusters. DBSCAN identified noise effectively but resulted in lower cluster compactness, with a Silhouette score of 0.55 and a Davies-Bouldin index of 1.50. The findings highlight Hierarchical Clustering as the most effective method for customer segmentation in banking, with flexibility depending on the data and objectives.</p>

2025-03-06T00:00:00+00:00 Copyright (c) 2025 Md Amran Hossen Pabel, Biswanath Bhattacharjee, Sonjoy Kumar Dey, Sakib Salam Jamee, Md Omar Obaid, Md Sakib Mia, Sajidul Islam Khan, Mohammad Kawsur Sharif