This paper will discuss about the different approaches to fraud detection such as Data mining, machine learning and artificial intelligence and statistical data analysis. Then we list some of the technical and troublesome challenges to modern fraud detection techniques. A comparison study of these techniques is also done according to the metrics like precision, False Alarm Rate (FAR), Accuracy, Cost, True Positive Rate (TPR) against different categories of frauds such as internal bank fraud, credit card fraud, loan fraud. Finally, we discuss the disadvantages of the existing fraud detection systems and we attempt to recommend a specific technique or algorithm for detecting a specific type of fraud with their advantages and disadvantages.
Copyrights © 2021