COST SENSITIVE FINANCIAL FRAUD DETECTION USING VALUE AT RISK AND MACHINE LEARNING MODELS

Abstract

In a cost-sensitive financial fraud detection system, our research uses advanced machine learning and Value at Risk (VaR) to help people perform more accurate financial uncertainty assessments. The vast majority of outdated fraud detection systems stress categorization over the financial consequences of false positives and negatives. The suggested approach uses a numerical risk indicator called Value at Risk to calculate the fraud costs. There are several parallels between financial risk management and predictive models. Cost-sensitive machine learning algorithms give high-risk agreements precedence over incorrect classifications in order to reduce expected financial losses. Supervised learning and statistical risk assessment enhance output and problem-solving skills. This makes it possible for organizations to minimize fraudulent activity and maximize their resources.