Journals / Turkish Journal of Science and Technology / 2026 / Volume: 21 Issue: 2

A Constraint-Aware Soft Voting Ensemble Framework for Credit Card Fraud Detection under Operational Threshold Optimization

Operasyonel Eşik Optimizasyonu Altında Kredi Kartı Dolandırıcılığı Tespiti için Kısıt Farkındalıklı Yumuşak Oylama Topluluk Çerçevesi

Büşra Demir Sezgin(İSTANBUL AYDIN ÜNİVERSİTESİ)0009-0005-3670-1911
Şükrü Mustafa Kaya(İSTANBUL AYDIN ÜNİVERSİTESİ)0000-0003-2710-0063
Pages
—
Publication Date
2026-09-30
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

These findings indicate that constraint-aware threshold optimization can improve the practical decision quality of ensemble-based fraud detection systems without requiring retraining of the underlying classifiers.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Credit card fraud detection is a challenging binary classification problem due to extreme class imbalance and the operational cost asymmetry between false negatives and false positives.
📊
Key Findings & Data:This study proposes a constraint-aware soft voting ensemble framework that explicitly separates model training from decision threshold selection.
💡
Conclusion & Impact:These findings indicate that constraint-aware threshold optimization can improve the practical decision quality of ensemble-based fraud detection systems without requiring retraining of the underlying classifiers.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Credit card fraud detection · ensemble learning · class imbalance

Abstract

Credit card fraud detection is a challenging binary classification problem due to extreme class imbalance and the operational cost asymmetry between false negatives and false positives. Although high-performing classifiers can provide strong probability estimates, their practical effectiveness depends heavily on how these probabilities are converted into final fraud decisions. This study proposes a constraint-aware soft voting ensemble framework that explicitly separates model training from decision threshold selection. The framework combines the probabilistic outputs of XGBoost, LightGBM, and Decision Tree classifiers through soft voting and then determines the final decision threshold on a validation set under predefined minimum recall and precision constraints. Unlike conventional approaches that rely on fixed probability thresholds, the proposed method aligns the classification decision with operational fraud detection requirements. Experimental results on a highly imbalanced credit card fraud dataset show that the proposed ensemble achieves a precision of 0.9000, recall of 0.8514, F1-score of 0.8750, ROC-AUC of 0.9525, and PR-AUC of 0.8441. The confusion matrix further demonstrates that the framework maintains a low false-positive rate while detecting most fraudulent transactions. These findings indicate that constraint-aware threshold optimization can improve the practical decision quality of ensemble-based fraud detection systems without requiring retraining of the underlying classifiers.

Özet

Kredi kartı dolandırıcılığı tespiti, aşırı sınıf dengesizliği ve hatalı negatifler ile hatalı pozitifler arasındaki operasyonel maliyet farklılıkları nedeniyle zorlu bir ikili sınıflandırma problemidir. Yüksek performanslı sınıflandırıcılar güçlü olasılık tahminleri üretebilse de, bu tahminlerin pratik başarımı büyük ölçüde nihai dolandırıcılık kararlarına nasıl dönüştürüldüğüne bağlıdır. Bu çalışmada, model eğitimi ile karar eşiği seçimini açık biçimde birbirinden ayıran kısıt farkındalıklı bir yumuşak oylama topluluk çerçevesi önerilmektedir. Önerilen çerçeve; XGBoost, LightGBM ve karar ağacı sınıflandırıcılarının olasılıksal çıktılarını yumuşak oylama yöntemiyle birleştirmekte ve nihai karar eşiğini doğrulama veri seti üzerinde önceden tanımlanmış minimum duyarlılık ve kesinlik kısıtları altında belirlemektedir. Sabit olasılık eşiklerine dayanan geleneksel yaklaşımlardan farklı olarak önerilen yöntem, sınıflandırma kararını dolandırıcılık tespit sistemlerinin operasyonel gereksinimleriyle uyumlu hale getirmektedir. Aşırı dengesiz kredi kartı dolandırıcılığı veri seti üzerinde gerçekleştirilen deneylerde önerilen topluluk modeli 0,9000 kesinlik, 0,8514 duyarlılık, 0,8750 F1-skoru, 0,9525 ROC-AUC ve 0,8441 PR-AUC değerlerine ulaşmıştır. Karmaşıklık matrisi sonuçları, çerçevenin düşük hatalı pozitif oranını korurken dolandırıcılık işlemlerinin büyük bölümünü tespit edebildiğini göstermektedir. Bulgular, kısıt farkındalıklı eşik optimizasyonunun, temel sınıflandırıcıların yeniden eğitilmesine gerek kalmadan topluluk tabanlı dolandırıcılık tespit sistemlerinin pratik karar kalitesini artırabileceğini göstermektedir.

🔖

Cite This Article

Sezgin, B. D., Kaya, Ş. M. (2026). A Constraint-Aware Soft Voting Ensemble Framework for Credit Card Fraud Detection under Operational Threshold Optimization. Turkish Journal of Science and Technology, Volume: 21 Issue: 2. https://doi.org/10.55525/tjst.1931327

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (24)

  1. Compagnino AA, Maruccia Y, Cavuoti S, Tutone A, Crupi R, Pagliaro A. An introduction to machine learning methods for fraud detection. Appl Sci 2025; 15: 1-32.
  2. Sun W, Shen Q, Gao Y, Mao Q, Qi T, Xu S. Objective over architecture: fraud detection under extreme imbalance in bank account opening. Computation 2025; 13: 1-17.
  3. Khalid AR, Owoh N, Uthmani O, Ashawa M, Osamor J, Adejoh J. Enhancing credit card fraud detection: an ensemble machine learning approach. Big Data and Cognitive Computing 2024; 8: 1-27.
  4. Ileberi E, Sun Y. A Hybrid Deep Learning Ensemble Model for Credit Card Fraud Detection. IEEE Access 2024; 12: 175829-175838.
  5. Zhao X, Liu Y, Zhao Q. Improved LightGBM for Extremely Imbalanced Data and Application to Credit Card Fraud Detection 2024; 12: 159316-159335.
  6. Ileberi E, Sun Y. Performance analysis of cost-sensitive oversampling mechanisms for credit card fraud detection. IEEE Access 2025; 13: 202655-202664.
  7. Iqbal S, Awan KM, Kamal S, Rehman ZU. Interpretable ensemble learning models for credit card fraud detection. Appl Sci 2025; 15: 1-30.
  8. Albalawi T, Dardouri S. Enhancing credit card fraud detection using traditional and deep learning models with class imbalance mitigation. Frontiers in Artificial Intelligence 2025, 8: 1-10.
  9. Siam AM, Bhowmik P, Uddin P. Hybrid feature selection framework for enhanced credit card fraud detection using machine learning models. PLoS One 2025, 20(7): 1-34.
  10. Mekhlouf HB, Moussaid A, Ghanimi F. Adaptive credit card fraud detection: reinforcement learning agents vs anomaly detection techniques. FinTech 2025; 5(1): 1-17.
  11. Hacini AD, Benabdelouahad M, Abassi I, Houhou S, Boulmerka A, Farhi N. LLM-assisted financial fraud detection with reinforcement learning. Algorithms 2025; 18: 1-27.
  12. Qian J, Tong G. Metapath-guided graph neural networks for financial fraud detection. Comput Electr Eng 2025; 126: 1-23.
  13. Lou C, Wang Y, Li J, Qian Y, Li X. Graph neural network for fraud detection via context encoding and adaptive aggregation. Expert Syst Appl 2025; 261: 1-12.
  14. Boyapati M, Aygun R. BalancerGNN: balancer graph neural networks for imbalanced datasets: a case study on fraud detection. Neural Netw 2025; 182: 1-14.
  15. Ntuala GM, Xia Q, Xia H, Badjie A, Mukala P, Tavares EEDS, Gao J, Ukwuoma CC. A dual-layer GNN with economic penalty mechanisms for blockchain fraud detection. Expert Syst Appl 2026; 299: 1-15.
  16. Olaniyan J, Olaniyan D, Obagbuwa IC, Ngafeeson M. Graph-temporal contrastive transformer for financial fraud detection using transaction behavior modeling. Algorithms 2025; 18: 1-23.
  17. Shao Z, Yu H, Wen J, Liu Z, Qi P. A graph fraud detection model based on mutual information. Neurocomputing 2026; 663: 1-11.
  18. Mrčela L, Kostanjčar Z. A hierarchical probabilistic deep learning approach for contextual anomaly detection in mixed-type tabular data. IEEE Access 2025; 13: 173298-173311.
  19. Almusallam N, Qayyum J. A hybrid feature selection and clustering-based ensemble learning approach for real-time fraud detection in financial transactions. Comput Mater Continua 2025; 85(2): 3653-3687.
  20. Chen Y, Zhang K, Zhu H, Qiu Z. A novel federated transfer learning framework for credit card fraud detection under heterogeneous data conditions. Risks 2025; 13: 1-24.
  21. Sampaio C, Silva R. Digital transformation in accounting: an assessment of automation and AI integration. Financial Stud 2025; 13(4): 1-23.
  22. Sarna NJ, Rithen FA, Jui US, Belal S, Amin A, Oishee TK, Islam AKMM. AI driven fraud detection models in financial networks: a comprehensive systematic review. IEEE Access 2025; 13: 141204-141233.
  23. Menezes RS, Filho RH. Investigating the robustness of graph neural networks to data drift: a case study on financial transaction data. IEEE Access 2025; 13: 164302-164312.
  24. Baisholan N, Dietz JE, Gnatyuk S, Turdalyuly M, Matson ET, Baisholanova K. A systematic review of machine learning in credit card fraud detection under original class imbalance. Computers 2025; 14: 1-23.