Detecting Account Takeover (ATO) in Fintech Companies Using Machine Learning

Financial technology usually called fintech, has changed the landscape of financial services by introducing technological innovations to simplify, digitize, and disrupt traditional services. However, because of these technological developments, account takeover (ATO) attacks are one of the new challenges that have appeared, threatening user security and platform reputation. Account Takeover (ATO) attacks cause a significant risk for many sectors, including financial services, online retail, and social media. However, most current efforts focus more on Sybil detection and fraud detection rather than ATO. ATO attacks cause significant harm to users and fintech companies, including identity loss and loss of customer trust. This study compares the effectiveness of Random Forest, Logistic Regression, XGBoost, and Light GBM in finding new fraud trends. This is achieved by applying Machine Learning (ML) methods to improve credit card fraud detection. This dataset contains 6,362,620 unbalanced data that are balanced using the undersampling technique. Among the four algorithms compared, the LightGBM algorithm achieves the best result with an accuracy of 0.99. These results enable fairly accurate fraud detection. This research contributes to fintech companies by advancing the understanding of ML techniques in fraud detection, providing a comparative analysis of different algorithms, and providing more effective strategies for solving credit card fraud problems.