LEVERAGING VALUE-AT-RISK AND MACHINE LEARNING FOR FINANCIAL FRAUD DETECTION IN IMBALANCED DATA
Keywords:
Financial Fraud Detection, Value-at-Risk (VaR), Machine Learning, Imbalanced Data, Anomaly Detection, CostSensitive Learning, Risk Management, Supervised Learning, Ensemble MethodsAbstract
It is difficult to detect financial crime due to the extremely skewed character of datasets including illicit transactions. When there are few instances, rule-based and statistical approaches may not be able to detect fraud tendencies. Improved fraud detection is the goal of this study, which integrates ML methods with the popular risk management metric Value-at-Risk (VaR). Value at Risk (VaR) provides a numerical assessment of a company's financial risk, which aids in the detection of scams. Sorting deals into groups and fixing class imbalances using cost-sensitive learning methods and resampling tactics are both accomplished by a multitude of machine learning algorithms. Among these, you can find anomaly detection algorithms, ensemble techniques, and supervised learning models. To demonstrate that the proposed strategy may enhance the precision and recall of fraud detection, it is tested on real-world financial datasets. The findings highlight the potential of financial risk assessment and data driven by artificial intelligence to combat financial crime.
