RIDIT scoring method for ranking finance fraud cases

The arising number of fraud cases poses a serious threat to financial security. Detecting fraud is challenging due to dataset characteristics, particularly the imbalance between fraudulent and non-fraudulent activities, which may lead to biased results. To address the issue, this study introduces the Relative to an Identified Distribution (RIDIT) scoring method to identify potential financial fraud. RIDIT scores are calculated for different years in financial data, helping rank financiers’ preferences based on fraud risk. The study applies this method to the New York Suspicious Activity Report Statistics dataset, which includes 14 types of fraudulent activities such as credit or debit card fraud, wire fraud, Ponzi schemes, and consumer loan fraud. Each category is ranked based on mean RIDIT scores, which range between 0 and 1. A score below 0.5 indicates lower fraud risk, while a score above 0.5 suggests higher potential fraud. Results show that credit or debit card fraud (0.3282) and wire fraud (0.4475) have the lowest potential fraud. In contrast, consumer loan fraud has the highest mean RIDIT score (0.7081), highlighting a greater risk and the need for close monitoring. The novelty of applying RIDIT Scoring in this study contributes to the financial system by providing a useful tool to rank suspicious activities. It helps the financiers and organizations to identify the high-risk of suspicious activity effectively. Thus, allows financiers minimize losses and promote financial transparency.