Developing a Financial Fraud Detection System Using Machine Learning Techniques

This study aimed to develop a Financial Fraud Detection System Using Machine Learning Techniques in Iraqi banks. To achieve this objective, an experimental approach was adopted, involving the creation of a database, simulation, data imbalance management, and evaluation of five main models. The results demonstrated the superiority of the random forest model, achieving a high F1 score (90.45%), a recall rate of 91.25%, and the fewest false alarms. The research concluded that the Random Forest-based system is a cost-effective solution for Iraqi banks and recommends its gradual adoption along with building a national infrastructure for sharing fraud data.