Sine Cosine based Zebra Optimization Algorithm with Support Vector Machine for Credit Card Fraud Detection

As credit card usage grows due to its efficiency and easy-to-use facilities, the risk of fraud and misuse has also risen. Processing the huge amount of data generated has become a challenge, leading to unbalanced and high-dimensional datasets, which impacts the performance of the model. Therefore, this research proposes a Sine Cosine-based Zebra Optimization Algorithm with Support Vector Machine (SCZOA-SVM) for classifying Credit Card Fraud Detection (CCFD). The adaptive search efficiency allows exploration in search space by capturing fine-grained patterns thereby SCZOA enhances robustness and accuracy in CCFD. Also, in SCZOA, the dynamic modifications in search parameters help to adapt in determining fraud patterns that enhance detection accuracy. The optimal combination of selected features is fed into the SVM classifier to classify data into fraud and normal transactions. Overall, the SCZOA-SVM achieves better performance in identifying fraudulent transactions by accomplishing the CCFD process effectively. The SCZOA-SVM achieves a better accuracy of ${9 9. 9 7 \%}$, recall/sensitivity of $s{9 8. 1 8 \%}$, precision of ${9 8. 2 7 \%}$, f1-score of ${9 8. 2 2 \%}$ and specificity of 99.97% which is better when compared to Support Vector Data Description (SVDD).