Machine learning models as tool for detecting suspicious bank operations

Abstract

Given the digital transformation of the banking system and the expanding availability of financial services, the problem of ensuring the security of banking operations becomes strategically important. The importance of the issue lies in studying the effectiveness of fraud prediction and prevention methods, which are key components of a complete economic security system in the banking sector.  Authors focus on the formation of bank card fraud detection models. The research proposes specific recommendations for improving anti-fraud systems in the banking sector using machine learning models. The researchers implemented and tested XGBoost and ANN models in Python to detect fraudulent transactions. The experiments substantiate their high performance and their flexibility and ability to adapt to new data. Using these models enabled banks for quick detection of suspicious operations, which reduces the risk of losses and improves the overall financial system security. Ensuring the secure functioning of the banking transaction system requires a comprehensive approach that includes not only the introduction of modern analytical tools but also continuous training and updating models to identify new methods of fraud. Implementation of the proposed measures will enable credit organizations to improve the effectiveness of fraud protection, reduce financial losses from fraudulent actions and strengthen client confidence.