LEVERAGING MACHINE LEARNING FOR PROACTIVE MONEY LAUNDERING DETECTION IN NIGERIA

This paper examined the effectiveness of machine learning (ML) Models for the ongoing war against money laundering in Nigeria. In advanced economies, the rise of machine learning has revolutionized the way financial institutions and governments combat illegal financial activities emanating from mobile money transactions. By using advanced algorithms and data analysis techniques, machine learning has proven to be an effective tool in identifying suspicious financial transactions and patterns, thereby helping authorities take proactive rather than reactive measures in preventing fraudulent transactions. This study aimed to address the research gap in the use of reactive approach by Nigeria government, where the prevalence of money laundering has risen in recent years, and explored how ML techniques can be utilized to enhance the country’s efforts in combating this financial crime. The datasets for this study were obtained from Kaggle website that contains fraudulent transactions on money laundering, which was used to train, validate and test the ML models. Logistic regression (LR) and Support Vector Machine (SVM) were used as the baseline models while Sparse Autoencoder (SAE) neural network was used for feature learning and dimensionality reduction. The results indicate that LR classifier still showed reasonable performance but did not outperform the other models. Among all the measures, SVM exhibited outstanding performance, with over 90% prediction accuracy. The amount of money transferred and location of transactions emerged as top features for predicting money laundering transactions in online money transfers. These findings suggest that further research is needed to enhance the logistic regression model, and sparse autoencoder neural network should be explored as potential tool for law enforcement agencies and Nigeria financial Institutions to proactively learn representative data from high dimensional datasets as quality of data improves performance of predictive models.