Decision Tree-Based Credit Card Fraud Detection System: Design and Optimization

Decision Tree-Based Credit Card Fraud Detection System: Design and Optimization

With the rapid development of fintech, credit card fraud has become increasingly sophisticated and intelligent, posing significant challenges to banks and consumers. Traditional fraud detection methods exhibit noticeable shortcomings in real-time performance, accuracy, and interpretability. This paper proposes an improved decision tree-based method for credit card fraud detection, enhancing detection performance by integrating dynamic feature engineering and ensemble learning strategies. The study employs a cost-sensitive decision tree algorithm as the base model, addressing class imbalance through sample weight adjustment, and designs a two-stage detection framework combined with random forest for result validation. Experimental results demonstrate that the proposed method significantly outperforms traditional approaches in both detection accuracy and false positive rate, particularly excelling in detecting emerging fraud patterns. The study validates the practical value of decision tree models in financial risk control and provides a highly interpretable, low-cost deployment solution for real-time anti-fraud systems.