The role of Louvain-coloring clustering in the detection of fraud transactions

Abstract

Clustering is a technique in data mining capable of grouping very large amounts of data to gain new knowledge based on unsupervised learning. Clustering is capable of grouping various types of data and fields. The process that requires this technique is in the business sector, especially banking. In the transaction business process in banking, fraud is often encountered in transactions. This raises interest in clustering data fraud in transactions. An algorithm is needed in the cluster, namely Louvain’s algorithm. Louvain’s algorithm is capable of clustering in large numbers, which represent them in a graph. So, the Louvain algorithm is optimized with colored graphs to facilitate research continuity in labeling. In this study, 33,491 non-fraud data were grouped, and 241 fraud transaction data were carried out. However, Louvain’s algorithm shows that clustering increases the amount of data fraud of 90% by accurate.