Identifying Potentially Illicit Money Laundering and Terrorism Financing Transactions Through Machine Learning Techniques

Financial institutions worldwide face significant challenges in detecting and preventing illicit financial activities, such as money laundering and terrorism financing. Traditional rule-based methods often generate high false positive rates, increasing manual verification efforts and higher operational costs. This research explores machine learning techniques to enhance the detection of suspicious transactions. Several algorithms, including K-Nearest Neighbors, Decision Tree, Random Forest, Logistic Regression, Support Vector Machine, and Naïve Bayes, are applied and evaluated using a dataset from a financial institution. After a comprehensive performance assessment, the Random Forest model is the most effective, exhibiting the highest accuracy of 0.9333 in identifying suspicious transactions while minimising false positives. These findings highlight the potential of integrating machine learning into financial crime prevention protocols. It serves as a guide for practitioners to predict suspicious transactions in financial institutions based on previous patterns of transactions. It also helps financial institutions to reduce compliance costs, which are typically higher than those of standard rule-based systems. However, this work presents a suspicious transaction prediction paradigm from prior behaviour with no transparency in features and with high accuracy and zero false positives, as with the Financial Action Task Force (FATF) promotion of new Anti-Money Laundering and Countering the Financing of Terrorism (AML/CFT) initiatives.