<![CDATA[Anomaly detection in financial transaction data is a crucial aspect due to the increasing use of e-money, which raises the risk of suspicious activities such as fraud and money laundering. This study applies the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm to cluster transaction data and identify anomalies based on three main variables: transaction amount, transaction frequency, and final balance. The optimal parameters were determined by evaluating various combinations of epsilon (ε) and minPts values using the Davies-Bouldin Index (DBI) as a clustering quality indicator. The analysis results indicate that the optimal parameters are ε of 0.2727 and minPts of 6, with a DBI score of 1.1753. DBSCAN successfully formed six main clusters and detected 138 data points as noise, indicating potentially abnormal transactions. These findings demonstrate that DBSCAN can effectively distinguish between normal and suspicious data without requiring prior assumptions on the number of clusters, contributing to the development of more accurate and adaptive digital transaction anomaly detection systems.]]>
Copyrights © 2025
