رتبه بندی اعتباری مشتریان حقیقی بانک مبتنی بر روشهای رگرسیون لجستیک و لجستیک MPLE و شبکه عصبی

نویسندگان
سید احمد عاملی
آرزو چقایی
حسین رضایی
گروه اقتصاد منابع و انرژی، دانشکده اقتصاد ،دانشگاه خوارزمی،تهران،ایران
چکیده
بانک ها می توانند با طراحی یک سیستم کارآمد مدیریت وام، کارایی را افزایش داده و احتمال عدم برگشت اصل و فرع وام کاهش دهند.. در این مقاله کارآیی مدل­های رگرسیون لجستیک، شبکه عصبی مصنوعی، به منظور پیش بینی ریسک اعتباری مشتریان حقیقی یا به عبارتی متقاضیان وام­های خرد که گروه زیادی از مشتریان نظام بانکی کشور را شامل می­شوند، مورد بررسی قرار گرفت. باتوجه به نامتعادل بودن تعداد داده­ها حدآستانه بهینه با بکارگیری دو منحنی درجه حساسیت و درجه تشخیص محاسبه شد و از این روش میزان ریسک اعتباری هریک از مدل­ها استخراج شد. در رگرسیون لجستیک برای برآورد ضرایب با توجه به تعداد اندک مشتریان بد­حساب بجای روش حداکثر درستنمایی از روش حداکثر درستنمایی تاواندیده استفاده شد. در نهایت میزان صحت و دقت هر مدل با معیارهای متعدد بررسی شد. با استفاده از منحنی راک به بررسی قدرت تفکیک کنندگی مدلها پرداخته که در اینجا مدل شبکه عصبی بهترین قدرت تفکیک کنندگی را دارا بود. سپس با مقایسه خطاهای MSE، RMSE و MAE کارایی سنجش روش­ها مورد مقایسه قرار گرفت و عملکرد لجستیک MPLE و شبکه عصبی تقریبا با یکدیگر یکسان است. و در نهایت با توجه به هدف بانک در سه سناریو حداقل ریسک اعتباری، تشخیص مشتریان خوش حساب و تفکیک مشتریان به ترتیب شبکه عصبی، لجستیکMPLE و در سناریوی سوم شبکه عصبی و لجستیکMPLE بطور همزمان به عنوان مدل های برتر انتخاب شده اند.
کلیدواژه‌ها
رتبه بندی اعتباری
شبکه عصبی
لجستیک
لجستیک MPLE
منحنی راک
داده کاوی

عنوان مقاله English

Credit Rating of Real Bank Customers Based on Logistic Regression and MPLE Logistic Methods and Neural Network

نویسنده English

Seyed Ahmad Ameli
چکیده English

By designing an efficient loan management system, banks can increase efficiency and reduce the probability of non-repayment of principal and sub-loans. In this paper, the efficiency of logistic regression models, artificial neural network, was examined to predict the credit risk of real customers or in other words, applicants for microloans, which include a large group of customers in the country's banking system. Given the imbalance of the number of data, the optimal threshold was calculated using two sensitivity and detection curves, and the credit risk of each model was extracted from this method. In logistic regression, the compensated maximum likelihood method was used to estimate the coefficients considering the small number of bad customers instead of the maximum likelihood method. Finally, the accuracy and precision of each model was examined with multiple criteria. Using the Rock curve, the resolution of the models was examined, where the neural network model had the best resolution. Then, by comparing the MSE, RMSE and MAE errors, the efficiency of the methods was compared, and the performance of MPLE logistics and neural network is almost the same. Finally, considering the bank's goal in three scenarios of minimum credit risk, identifying good customers and separating customers, neural network, MPLE logistics, and in the third scenario, neural network and MPLE logistics simultaneously have been selected as the best models.

کلیدواژه‌ها English

Credit Rating
Neural Network
Logistics
MPLE logistics
Rock curve
Data mining
15. Al Janabi, M.A.M. (2021), "Multivariate portfolio optimization under illiquid market prospects: a review of theoretical algorithms and practical techniques for liquidity risk management", Journal of Modelling in Management, Vol. 16 No. 1, pp. 288-309. https://doi.org/10.1108/JM2-07-2019-0178
16. Baesens, B., Rösch, D., & Scheule, H. (2021). Credit Risk Analytics. Wiley.
17. Basel Committee on Banking Supervision. (2019). Credit risk: IRB approach. Bank for International Settlements.
18. Challa, K. (2024). Enhancing credit risk assessment using AI and big data in modern finance. American Data Science Journal for Advanced Computations (ADSJAC) ISSN: 3067-4166, 2(1)
19. Harb, E., El Khoury, R., Mansour, N. and Daou, R. (2023), "Risk management and bank performance: evidence from the MENA region", Journal of Financial Reporting and Accounting, Vol. 21 No. 5, pp. 974-998. https://doi.org/10.1108/JFRA-07-2021-0189
20. Harrell, FE Jr. (2001); Regression modeling strategies, New York: Springer Verlag, Berlin.
21. Haykin, Simon. (1999); Neural Networks, Macmillan College Publishing Company.https://doi.org/10.30574/gscarr.2024.18.2.0077
22. Kim, S., & Lee, H. (2023). A hybrid approach to credit risk evaluation using neural networks and logistic regression. Journal of Risk Finance, 24(1), 45-60
23. King, Gary., Zeng, Langche. (2001); “Logistic Regression in Rare Events Data” Political Analysis , VOL. 9, NO. 2, PP. 137-163.
24. Korsholm, Lars. (2004); “Analysis Of Diagnostic Studies, Sensitivity and specificity positive and negative predicted values ROC curves tests based on logistic regression”, Department Of Statistics And Demography, University Of Southern Denmark
25. kuhn, max., johnson, kjell. (2013); applied predictive modeling, springer.
26. Le, T., & Vo, H. (2024). Machine learning applications in banking: A focus on credit scoring. Global Finance Journal, 11(2), 101-115.
27. Louzada, F., Ara, A., & Fernandes, G. (2020). Classification methods applied to credit scoring: A systematic review. Expert Systems with Applications, 142.
28. Metz,CE,.Herman,BA.,Shen, JH.,(1998);“Maximum likelihood estimation of receiver operating characteristic (ROC) curves from continuously distributed data”, VOL.15, NO. 17, PP. 1033–1053.
29. Nguyen, P., & Tran, D. (2023). Logistic regression versus MPLE in predicting customer creditworthiness. Banking & Finance Review, 12(4), 89-102.
30. Nikolic, N., Zarkic-Joksimovic, N., Stojanovski, D., & Joksimovic, I. (2013). The application of brute force logistic regression to corporate credit scoring models: Evidence from Serbian financial statements. Expert Systems with Applications, 40(15), 5932-5944.‌
31. Patel, R., & Sharma, N. (2024). Comparative analysis of credit scoring methods: Neural networks and logistic regression. Finance and Economics Review, 18(1), 75-91.
32. Pavlou, A., & Dimitriou, N. (2020). Handling imbalanced data in credit scoring. Journal of Risk Model Validation.
33. Predictive analytics in credit risk management for banks: A comprehensive review-
34. Reza, M. S., Mahmud, M. I., Abeer, I. A., & Ahmed, N. (2024). Linear Discriminant Analysis in Credit Scoring: A Transparent Hybrid Model Approach. arXiv preprint arXiv:2412.04183.‌
35. Roberts, T., & Evans, C. (2024). Leveraging neural networks for effective credit scoring. Journal of Financial Engineering, 14(1), 35-50.
36. Saunders. Anthony,, , Allen, Linda. (2002); Credit Risk Measurement, Second Edition, New York, John Wiley & Sons, Inc.
37. Shah, S.A.A., Sukmana, R. and Fianto, B.A. (2020), "Duration model for maturity gap risk management in Islamic banks", Journal of Modelling in Management, Vol. 15 No. 3, pp. 1167-1186. https://doi.org/10.1108/JM2-08-2019-0184
38. Smith, J., & Brown, A. (2022). The effectiveness of logistic regression in credit scoring models. Journal of Financial Analytics, 15(2), 123-145
39. Spuchakova, E. and Cug, J. (2115). Credit Risk and LGD Modelling. Journal o of Procedia Economics and Finance, Vol. 23, pp. 433–444. : http://isiarticles.com/article/46621
40. Stiglitz, J. E., & Weiss, A. (1981). Credit Rationing in Markets with Imperfect Information. American Economic Review, 71(3), 393-410
41. Topyan, Kudret, Chia-Jane Wang, Natalia Boliari, and Carlos Elias. (2024). Credit Risk Management and US Bank-Holding Companies: An Empirical Investigation. Journal of Risk and Financial Management 17: 56. https://doi.org/10.3390/ jrfm17020056
42. Turan, H.(2016)," The weighting of factors affecting credit risk in banking. Istanbul Conference of Economics and Finance, ICEF 2115, 22-23 October 2115, Istanbul, Turkey. Retrived at: https://www.researchgate.net/publication/304000721
43. Wang, L., & Zhao, Q. (2022). Enhancing credit risk assessment using MPLE logistic regression. International Journal of Banking Studies, 10(1), 56-72.
44. Xia, Y., Liu, C., Li, Y., & Liu, N. (2018). A boosted decision tree approach using Bayesian hyper-parameter optimization for credit scoring. Expert Systems with Applications.
45. Zhao, F. (2025). Evaluating Machine Learning Models for Personal Credit Risk: A Comparison of Logistic Regression, Random Forest, and XGBoost. Applied and Computational Engineering, 158, 232-241. https://doi.org/10.54254/2755-2721/2025.TJ23486