Identification of Suspicious E-Banking Transactions Pattern

The volume of data of banks has increased over time due to providing various services. On the other hand, such growth and development have paved the new grounds for frauds and offenses. Therefore, fraud detection in the different fields of electronic banking has converted into a fundamental challenge and various methods have been employed for this purpose. Nowadays, procedures and thus fraud detections have become far more complicated with the technology developing and with the business processes becoming more complicated. This research could develop an identification model for suspicious transaction patterns on datasets including the data of transactions of one of the Iranian domestic banks through data mining techniques. It focused on optimizing model results, achieving maximum precision and recall in identification of suspicious transactions, and improving model reliability. RapidMiner was used as a tool for preparing data and modeling to realize research objectives.

Keywords:

fraud , identification of suspicious transactions , decision tree , optimization , model reliability