Publication

Abstract : In the era of digital financial transactions, credit card fraudemerges as a significant threat, demanding advanced detection methodologies. This paper presents an efficient and effective solution using Light Gradient Boosting Machine, a cutting edge machine learning technique. Our approach meticulously preprocesses a dataset of anonymized credit card transactions, emphasizing normalization, feature selection, and strategic data splitting to address class imbalance. The Light Gradient Boosting Machine model, optimized through Randomized Search Cross-Validation, fine-tunes key parameters to enhance predictive accuracy while mitigating overfitting. The model’s performance, evaluated on various metrics such as accuracy, precision, recall, F1-score, and Area Under the Receiver Operating Characteristic Curve exhibits high precision, ensuring minimal false positives, and substantial recall for effective fraud detection. With an impressive Area Under the Curve score, our model demonstrates superior discriminative capabilities. The findings of this study underscore the potential of Light Gradient Boosting Machine in credit card fraud detection, offering a scalable, reliable tool for integration into financial security systems.