BANKACILIK İÇ DENETİM SÜREÇLERİNDE RİSK ODAKLI KARAR DESTEK MODELİ GELİŞTİRİLMESİ

Öz

Bankacılık sektöründe iç denetim faaliyetleri, sınırlı kaynakların etkin kullanımını gerektirmekte ve denetçilerin hangi kredi segmentlerini önceliklendirmesi gerektiği kritik bir karar problemi oluşturmaktadır. Bu çalışmada, bankacılık iç denetimi için karar destek amaçlı iki seviyeli hibrit bir sistem geliştirilmiştir. Çalışmada LendingClub platformundan elde edilen ve 1.417.571 kredi kaydı içeren veri seti, BDDK terminolojisi, TCMB döviz kurları ve Türkiye makroekonomik göstergeleri dikkate alınarak Türkiye bankacılık standartlarına uyarlanmıştır. Birinci seviyede, Çok Kriterli Karar Verme (ÇKKV) yöntemlerinden Analitik Hiyerarşi Süreci (AHP), TOPSIS ve VIKOR teknikleri kullanılarak 14 kredi segmenti arasından stratejik denetim alanı seçimi gerçekleştirilmiştir. AHP ile subjektif, Entropy yöntemi ile objektif kriter ağırlıkları hesaplanarak hibrit ağırlıklandırma yaklaşımı uygulanmış ve en yüksek TOPSIS skoru (0,7202) ile Borç Birleştirme segmenti öncelikli denetim alanı olarak belirlenmiştir. İkinci seviyede, 778.684 kayıt içeren seçili segment üzerinde XGBoost, Random Forest ve Lojistik Regresyon modelleri GridSearchCV ile optimize edilmiş, en iyi performansı gösteren XGBoost modeli (AUC-ROC: 0,7665) ile Isolation Forest anomali tespiti birleştirilerek hibrit skor füzyonu oluşturulmuştur. Optimal füzyon ağırlıkları %95 denetimli + %5 denetimsiz olarak belirlenmiştir. Örneklem boyutu Cochran formülü ile 133 olarak hesaplanmış, %50 yüksek riskli + %50 temsilci dosya stratejisiyle hibrit örnekleme gerçekleştirilmiş ve SHAP analizi ile model açıklanabilirliği sağlanmıştır. Walk-Forward validasyon ile 2018 yılı verileri üzerinde zamansal performans test edilmiştir. Sonuç olarak, önerilen hibrit sistem, iç denetim süreçlerinde denetçi deneyimine dayalı kararları veri odaklı bir yaklaşıma dönüştürerek rastgele örneklemeye kıyasla kontrollü tek örneklem karşılaştırmasında belirgin biçimde daha yüksek bir risk yakalama kapasitesi sergilemiştir ve banka iç denetim birimlerinin kaynak verimliliğini artırma potansiyelini ortaya koymuştur.

Anahtar Kelimeler

• Çok Kriterli Karar Verme
• Makine Öğrenmesi
• Bankacılık İç Denetimi
• Hibrit Sistem
• Risk Tabanlı Örnekleme

DEVELOPMENT OF A RISK-BASED DECISION SUPPORT MODEL FOR BANKING INTERNAL AUDIT PROCESSES

Öz

Internal audit activities in the banking sector require the effective utilization of limited resources, and determining which credit segments auditors should prioritize constitutes a critical decision problem. In this study, a two-level hybrid decision support system has been developed for banking internal audit purposes. The dataset obtained from LendingClub platform, containing 1.417.571 credit records, was adapted to Turkish banking standards considering BRSA terminology, CBRT exchange rates, and Turkish macroeconomic indicators. At the first level, strategic audit area selection was performed among 14 credit segments using Multi-Criteria Decision Making (MCDM) methods including Analytic Hierarchy Process (AHP), TOPSIS, and VIKOR techniques. A hybrid weighting approach was applied by calculating subjective weights through AHP and objective weights through Entropy method, and Debt Consolidation segment was identified as the priority audit area with the highest TOPSIS score (0.7202). At the second level, XGBoost, Random Forest, and Logistic Regression models were optimized using GridSearchCV on the selected segment containing 778,684 records, and hybrid score fusion was created by combining the best-performing XGBoost model (AUC-ROC: 0.7665) with Isolation Forest anomaly detection. Optimal fusion weights were determined as 95% supervised + 5% unsupervised. Sample size was calculated as 133 using the Cochran formula, hybrid sampling was performed with 50% high-risk + 50% representative files strategy, and model interpretability was ensured through SHAP analysis. Temporal performance was tested on 2018 data using Walk-Forward validation. In conclusion, the proposed hybrid system has demonstrated a notably higher risk detection capacity compared to random sampling in a controlled single-sample comparison by transforming auditor experience-based decision-making into a data-driven approach in internal audit processes, thereby revealing its potential to enhance resource efficiency in bank internal audit units.

Anahtar Kelimeler

• Multi-Criteria Decision Making
• Machine Learning
• Banking Internal Audit
• Hybrid System
• Risk-Based Sampling

Kaynakça

1. Ahmed, M., Mahmood, A. N., and Islam, M. R. (2016). A survey of anomaly detection techniques in financial domain. Future Generation Computer Systems, 55, 278-288.
2. AICPA. (2014). Audit Guide: Audit Sampling. New York: American Institute of Certified Public Accountants, 45-78.
3. Arrieta, A. B., Díaz-Rodríguez, N., Del Ser, J., Bennetot, A., Tabik, S., Barbado, A., Garcia, S., Gil-Lopez, S., Molina, D., Benjamins, R., Chatila, R., and Herrera, F. (2020). Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI. Information Fusion, 58, 82-115.
4. Bankacılık Düzenleme ve Denetleme Kurumu. (2014). Bankaların İç Sistemleri ve İçsel Sermaye Yeterliliği Değerlendirme Süreci Hakkında Yönetmelik. Ankara: BDDK Yayınları.
5. Basel Committee on Banking Supervision. (2012). The internal audit function in banks. Basel: Bank for International Settlements, 1-28.
6. Basel II: International Convergence of Capital Measurement and Capital Standards. (2004). Basel: Bank for International Settlements.
7. Basel III: A Global Regulatory Framework for More Resilient Banks and Banking Systems. (2010). Basel: Bank for International Settlements.
8. Breiman, L. (2001). Random forests. Machine Learning, 45(1), 5-32.

9. Chen, T., and Guestrin, C. (2016). XGBoost: A Scalable Tree Boosting System. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Francisco, 785-794.
10. Chen, T., and Guestrin, C. (2020). XGBoost: Extreme Gradient Boosting (Version 1.7.0) [Computer Software]. URL: https://github.com/dmlc/xgboost
11. Chen, Y., Li, W., and Zhang, Y. (2018). Explaining credit risk prediction with SHAP values. 2018 IEEE International Conference on Big Data, Seattle, 4539-4544.
12. Cochran, W. G. (1977). Sampling Techniques (Third Edition). New York: John Wiley & Sons, 72-88.
13. Dietterich, T. G. (2000). Ensemble methods in machine learning. International Workshop on Multiple Classifier Systems, Cagliari, 1-15.
14. Griffiths, D. (2005). Risk Based Internal Auditing: An Introduction. London: The Institute of Internal Auditors UK and Ireland, 12-34.
15. Hand, D. J. (2006). Classifier technology and the illusion of progress. Statistical Science, 21(1), 1-14.
16. Hosmer, D. W., and Lemeshow, S. (2000). Applied Logistic Regression (Second Edition). New York: John Wiley & Sons, 156-198.
17. Hwang, C. L., and Yoon, K. (1981). Multiple Attribute Decision Making: Methods and Applications - A State-of-the-Art Survey. New York: Springer-Verlag, 58-191.
18. İnternet: Kaggle. (2024). LendingClub Loan Data. URL: https://www.kaggle.com/wendykan/lending-club-loan-data, Son Erişim Tarihi: 15.09.2024.
19. Institute of Internal Auditors. (2017). International Professional Practices Framework (IPPF). Florida: The Institute of Internal Auditors, 1-156.
20. Jans, M., Alles, M., and Vasarhelyi, M. (2014). A Field Study on the Use of Process Mining of Event Logs as an Analytical Procedure in Auditing. The Accounting Review, 89(5), 1751-1773.
21. Lessmann, S., Baesens, B., Seow, H. V., and Thomas, L. C. (2015). Benchmarking state-of-the-art classification algorithms for credit scoring: An update of research. European Journal of Operational Research, 247(1), 124-136.
22. Liu, F. T., Ting, K. M., and Zhou, Z. H. (2008). Isolation forest. Proceedings of the 8th IEEE International Conference on Data Mining, Pisa, 413-422.
23. Lundberg, S. M., and Lee, S. I. (2017). A Unified Approach to Interpreting Model Predictions. Advances in Neural Information Processing Systems 30, Long Beach, 4765-4774.
24. Lundberg, S. M. (2021). SHAP: SHapley Additive exPlanations (Version 0.41.0) [Computer Software]. URL: https://github.com/slundberg/shap
25. Ma, X., Sha, J., Wang, D., Yu, Y., Yang, Q., and Niu, X. (2018). Study on a prediction of P2P network loan default based on the machine learning LightGBM and XGBoost algorithms according to different high dimensional data cleaning. Electronic Commerce Research and Applications, 31, 24-39.
26. MacQueen, J. (1967). Some methods for classification and analysis of multivariate observations. Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability, Berkeley, 1, 281-297.
27. NumPy Developers. (2020). NumPy (Version 1.21.0) [Computer Software]. URL: https://numpy.org
28. Opricovic, S., and Tzeng, G. H. (2004). Compromise solution by MCDM methods: A comparative analysis of VIKOR and TOPSIS. European Journal of Operational Research, 156(2), 445-455.
29. Pandas Development Team. (2020). pandas: Powerful Python Data Analysis Toolkit (Version 1.3.0) [Computer Software]. URL: https://pandas.pydata.org
30. Pickett, K. H. S. (2010). The Internal Auditing Handbook (Third Edition). West Sussex: John Wiley & Sons, 189-234.
31. Saaty, T. L. (1980). The Analytic Hierarchy Process. New York: McGraw-Hill, 1-287.
32. Saaty, T. L. (1990). How to make a decision: The analytic hierarchy process. European Journal of Operational Research, 48(1), 9-26.
33. Scikit-learn Developers. (2021). scikit-learn: Machine Learning in Python (Version 1.0.0) [Computer Software]. URL: https://scikit-learn.org
34. Serrano-Cinca, C., Gutiérrez-Nieto, B., and López-Palacios, L. (2015). Determinants of default in P2P lending. PloS One, 10(10), e0139427.
35. Shannon, C. E. (1948). A Mathematical Theory of Communication. The Bell System Technical Journal, 27(3), 379-423.
36. Shen, K. Y., Yan, M. R., and Tzeng, G. H. (2020). Combining MCDM and machine learning approaches for financial risk evaluation. Applied Soft Computing, 94, 106436.
37. Türkiye Bankalar Birliği. (2020). Türkiye'de Bankacılık Sistemi Seçilmiş Rasyolar 2020. İstanbul: TBB Yayınları.
38. Türkiye Cumhuriyet Merkez Bankası. (2018). Yıllık Ortalama Döviz Kurları. Ankara: TCMB.
39. Türkiye İstatistik Kurumu. (2018). Satın Alma Gücü Paritesi Hesaplamaları. Ankara: TÜİK Yayınları.
40. Zhang, Y., Liu, G., and Jin, Z. (2020). Cluster-based sampling for financial auditing. Decision Support Systems, 134, 113301.
41. Zou, Z. H., Yun, Y., and Sun, J. N. (2006). Entropy method for determination of weight of evaluating indicators in fuzzy synthetic evaluation for water quality assessment. Journal of Southeast University (English Edition), 22(2), 266-269.
42. 5411 Sayılı Bankacılık Kanunu. (2005). T.C. Resmi Gazete, 25983, 1 Kasım 2005.