Fraud Detection Decision Support System for Indonesian Financial Institution

With the rise of technology, the channel of transaction increases, which increases the amount of transactions and opportunity for fraudsters to commit fraudulent acts. The current fraud detection system applied in financial institutions relies heavily on manual intervention, consuming a lot of time and prone to errors. Hence, a fraud detection decision support system will be created to aid financial institutions to detect suspicious activities quickly and requires minimum manual intervention. Classification rule will be set up depending on fraud scenario based on real-life use cases. Common scenarios will be adopted into learning models for the dataset to be processed. Transactional dataset will be prepared according to the predefined template and scenario. The result of the processing will be categorized based on the Risk Level which was derived from the Confidence Level that were obtained. Testing were conducted using 3 different datasets using a set of parameters. The result shows that the current algorithm detects all suspicious activity that were inserted into the financial dataset.