Identifying Suspicious Financial Transaction Using Machine Learning and Phytagorean Tree

Suspicious Financial Transaction Detection (SFTD) is a vital process in recognizing threats and fraudulent transactions. SFTD involves tracking historical and real-time transaction data, posing challenges in identifying true suspicious transactions. Although a rule-based system is used at the Regional Development Bank in Indonesia, its effectiveness remains limited due to the growing customer transactions. This study aims to address this issue using machine learning to reduce false positives while maintaining a manageable false-negative rate. Money laundering stages-placement, layering, integration-are explored, emphasizing the need to integrate illicit funds into legitimate financial systems. The Indonesian anti-money laundering process is outlined, involving alert generation, manual and financial institution investigations. Earlier research compared machine learning algorithms and found Random Forest outperforms other model to exhibit the highest accuracy. The study aims to employ a machine learning model on actual customer transaction data to enhance detection and prediction of suspicious transaction, such as money laundering, potentially streamlining manual efforts and achieving research objectives.