Real Time Credit Card Transaction Classification System Using High Dimension Feature Optimization

Credit card fraud is a significant challenge in financial transactions. Traditional rule-based detection methods fail to adapt to evolving fraud techniques. This paper explores machine learning (ML) models, including logistic regression, decision trees, random forests, and deep learning, to improve fraud detection accuracy. The proposed system integrates anomaly detection and ensemble learning to reduce false positives and enhance predictive performance. The research highlights data preprocessing techniques, feature selection strategies, and model evaluation metrics to develop a robust fraud detection system. Experimental results demonstrate an increased fraud detection rate while minimizing legitimate transaction disruptions. Additionally, real- world dataset testing provides insights into the advantages and limitations of different ML techniques.