An Explainable Hybrid Machine Learning Framework for Loan Approval Prediction Based on Random Forest, Light Gradient Boosting Machine, Combined with Grey Wolf Optimizer Algorithms

Öz

Loan approval prediction remains one of the most vital problems in financial decision-making at financial institutions. In recent years, these institutions have seen an increase in credit applications. For these reasons, loan approval prediction has become central to research and practice in credit risk management. This paper presents an explainable hybrid ML approach to predict the loan approval or rejection. The whole process for building the ML model is based on the well-known CRISP-DM process model. A publicly available dataset to predict the loan approval or rejection from Kaggle is used for this study. The raw dataset is first preprocessed to handling outliers, to transform the categorical variables into numerical variables, and then to perform the Min-Max normalization. In the experimental result section, nine different classification algorithms, namely Decision Trees, Random Forest, Gradient Boosting Machines, XGBoost, LightGBM, CatBoost, Support Vector Machines, Naive Bayes, and K-Nearest Neighbors, are employed to produce the prediction performance. To improve the prediction performance of the established methods, the optimized model parameters were obtained by Grey Wolf Optimizer (GWO), and the prediction performance was evaluated by the 3-fold cross-validation (CV). According to the obtained results, state-of-the-art ensemble-based algorithms clearly outperformed the more conventional classifier. It has been observed that the GWO-optimized LightGBM model achieved the highest cross-validation accuracy of 92.17%. Due to the majority of the ensemble methods being black-box, SHAP-Explainable Artificial Intelligence (XAI) was utilized to gain insight into the black-box decision-making process of the models. The results show that advanced ensemble learning techniques, metaheuristic-based hyperparameter optimization, and explainable AI can be combined to form a single coherent pipeline.

Anahtar Kelimeler

• Grey Wolf Optimizer
• hyperparameter optimization
• LightGBM
• loan prediction
• Random Forest Algorithm
• XAI

Destekleyen Kurum

The author(s) received no financial support from any institution, organization, or funding agency for the research, authorship, and/or publication of this article.

Proje Numarası

Not Exist

Etik Beyan

This study used a publicly available secondary dataset obtained from Kaggle. Since the study did not involve direct interaction with human participants, experimental intervention, or the collection of personally identifiable information, institutional ethics committee approval was not required. The authors declare that the study was conducted in accordance with the principles of research integrity, transparency, and publication ethics.

Teşekkür

The authors would like to express their profound gratitude to Prof. Dr. Derviş Karaboğa for his invaluable encouragement, inspiring guidance, and generous support regarding the conference submission. The authors also sincerely acknowledge Prof. Dr. Ayşegül Alaybeyoğlu for her continuous support, constructive guidance, and distinguished leadership throughout the entire process.

Kredi Onayı Tahmini için Rastgele Orman Tabanlı Nitelik Seçimi, LightGBM ve Gri Kurt Optimizasyonuna Dayalı Açıklanabilir Hibrit Makine Öğrenmesi Çerçevesi

Öz

Kredi onayı tahmini, finansal kurumlarda finansal karar verme süreçlerinin en önemli problemlerinden biri olmayı sürdürmektedir. Son yıllarda bu kurumlar kredi başvurularında artış gözlemlemiştir. Bu nedenlerle, kredi onayı tahmini kredi riski yönetiminde hem araştırma hem de uygulama açısından merkezi bir konu hâline gelmiştir. Bu çalışma, kredi onayı veya reddini tahmin etmek amacıyla açıklanabilir hibrit bir makine öğrenmesi yaklaşımı sunmaktadır. Makine öğrenmesi modelinin oluşturulmasına ilişkin tüm süreç, yaygın olarak bilinen CRISP-DM süreç modeline dayanmaktadır. Bu çalışmada, kredi onayı veya reddini tahmin etmek amacıyla Kaggle’dan elde edilen halka açık bir veri seti kullanılmıştır. Ham veri seti öncelikle aykırı değerlerin ele alınması, kategorik değişkenlerin sayısal değişkenlere dönüştürülmesi ve ardından Min-Max normalizasyonunun uygulanması için ön işleme tabi tutulmuştur. Deneysel sonuçlar bölümünde, tahmin performansını ortaya koymak amacıyla Karar Ağaçları, Rastgele Orman, Gradient Boosting Machines, XGBoost, LightGBM, CatBoost, Destek Vektör Makineleri, Naive Bayes ve K-En Yakın Komşu olmak üzere dokuz farklı sınıflandırma algoritması kullanılmıştır. Kurulan yöntemlerin tahmin performansını artırmak amacıyla optimize edilmiş model parametreleri Gri Kurt Optimizasyon Algoritması (GWO) ile elde edilmiş ve tahmin performansı 3 katlı çapraz doğrulama yöntemiyle değerlendirilmiştir. Elde edilen sonuçlara göre, güncel topluluk tabanlı algoritmalar daha geleneksel sınıflandırıcıları açıkça geride bırakmıştır. GWO ile optimize edilmiş LightGBM modelinin %92,17 ile en yüksek çapraz doğrulama doğruluğuna ulaştığı gözlemlenmiştir. Topluluk yöntemlerinin büyük çoğunluğunun kara kutu niteliğinde olması nedeniyle, modellerin kara kutu karar verme süreçlerine ilişkin içgörü elde etmek amacıyla SHAP tabanlı Açıklanabilir Yapay Zeka (XAI) kullanılmıştır. Bulgular, gelişmiş topluluk öğrenmesi tekniklerinin, meta-sezgisel tabanlı hiperparametre optimizasyonunun ve açıklanabilir yapay zekanın tek ve tutarlı bir işlem hattı içinde birleştirilebileceğini göstermektedir.

Anahtar Kelimeler

• Açıklanabilir Yapay Zeka
• Gri Kurt Optimizasyonu
• Hiperparametre Optimizasyonu
• Kredi Tahmini
• LightGBM
• Rastgele Orman Algoritması

Destekleyen Kurum

Bulunmamaktadır.

Proje Numarası

Not Exist

Etik Beyan

Bu çalışmada kullanılan veri seti halka açık erişime sahip ikincil bir veri kaynağından elde edilmiştir. Çalışma kapsamında insan katılımcılardan doğrudan veri toplanmamış, herhangi bir deneysel müdahale yapılmamış ve kişisel/veri gizliliğini ihlal edecek nitelikte tanımlayıcı bilgi kullanılmamıştır. Bu nedenle çalışma etik kurul izni gerektirmemektedir. Çalışmanın tüm aşamalarında bilimsel araştırma ve yayın etiği ilkelerine uygun hareket edilmiştir.

Teşekkür

Yazarlar, konferans başvurusu sürecindeki kıymetli teşvikleri, ilham verici akademik vizyonu, yol gösterici yaklaşımı ve cömert destekleri için Prof. Dr. Derviş Karaboğa’ya en derin şükranlarını ve saygılarını sunar. Ayrıca, tüm süreç boyunca göstermiş olduğu sürekli destek, yapıcı yönlendirme, üstün katkı ve seçkin liderliği için Prof. Dr. Ayşegül Alaybeyoğlu’na içten teşekkürlerini ve saygılarını sunar.

Kaynakça

1. Dastile X, Celik T, Potsane M. Statistical and machine learning models in credit scoring: A systematic literature survey. Applied Soft Computing 2020;91:106263. https://doi.org/10.1016/j.asoc.2020.106263.
2. Hand DJ, Henley WE. Statistical Classification Methods in Consumer Credit Scoring: A Review. Journal of the Royal Statistical Society Series A (Statistics in Society) 1997;160:523–41.
3. Lessmann S, Baesens B, Seow H-V, Thomas LC. Benchmarking state-of-the-art classification algorithms for credit scoring: An update of research. European Journal of Operational Research 2015;247:124–36. https://doi.org/10.1016/j.ejor.2015.05.030.
4. Thomas L, Crook J, Edelman D. Credit Scoring and Its Applications, Second Edition. Philadelphia, PA: Society for Industrial and Applied Mathematics; 2017. https://doi.org/10.1137/1.9781611974560.
5. Yang L, Shami A. On hyperparameter optimization of machine learning algorithms: Theory and practice. Neurocomputing 2020;415:295–316. https://doi.org/10.1016/j.neucom.2020.07.061.
6. Dastile X, Celik T, Potsane M. Statistical and machine learning models in credit scoring: A systematic literature survey. Applied Soft Computing 2020;91:106263. https://doi.org/10.1016/j.asoc.2020.106263.
7. Ayari H, Guetari PrR, Kraïem PrN. Machine learning powered financial credit scoring: a systematic literature review. Artif Intell Rev 2025;59:13. https://doi.org/10.1007/s10462-025-11416-2.
8. Noriega JP, Rivera LA, Herrera JA. Machine Learning for Credit Risk Prediction: A Systematic Literature Review. Data 2023;8:169. https://doi.org/10.3390/data8110169.

9. Liu Y, Yang M, Wang Y, Li Y, Xiong T, Li A. Applying machine learning algorithms to predict default probability in the online credit market: Evidence from China. International Review of Financial Analysis 2022;79:101971. https://doi.org/10.1016/j.irfa.2021.101971.
10. Chang V, Sivakulasingam S, Wang H, Wong ST, Ganatra MA, Luo J. Credit Risk Prediction Using Machine Learning and Deep Learning: A Study on Credit Card Customers. Risks 2024;12:174. https://doi.org/10.3390/risks12110174.
11. Zhang X, Zhang T, Hou L, Liu X, Guo Z, Tian Y, et al. Data-Driven Loan Default Prediction: A Machine Learning Approach for Enhancing Business Process Management. Systems 2025;13:581. https://doi.org/10.3390/systems13070581.
12. Bhandary R, Ghosh BK. Credit Card Default Prediction: An Empirical Analysis on Predictive Performance Using Statistical and Machine Learning Methods. JRFM 2025;18:23. https://doi.org/10.3390/jrfm18010023.
13. Fitzpatrick T, Mues C. An empirical comparison of classification algorithms for mortgage default prediction: evidence from a distressed mortgage market. European Journal of Operational Research 2016;249:427–39. https://doi.org/10.1016/j.ejor.2015.09.014.
14. Gambacorta L, Huang Y, Qiu H, Wang J. How do machine learning and non-traditional data affect credit scoring? New evidence from a Chinese fintech firm. Journal of Financial Stability 2024;73:101284. https://doi.org/10.1016/j.jfs.2024.101284.
15. Elmasry NH, Elshaarawy MK. Hybrid metaheuristic optimized Catboost models for construction cost estimation of concrete solid slabs. Sci Rep 2025;15:21612. https://doi.org/10.1038/s41598-025-06380-4.
16. Černevičienė J, Kabašinskas A. Explainable artificial intelligence (XAI) in finance: a systematic literature review. Artif Intell Rev 2024;57:216. https://doi.org/10.1007/s10462-024-10854-8.
17. Nallakaruppan MK, Balusamy B, Shri ML, Malathi V, Bhattacharyya S. An Explainable AI framework for credit evaluation and analysis. Applied Soft Computing 2024;153:111307. https://doi.org/10.1016/j.asoc.2024.111307.
18. Li H, Wu W. Loan default predictability with explainable machine learning. Finance Research Letters 2024;60:104867. https://doi.org/10.1016/j.frl.2023.104867.
19. Xie S, Shingadia T. Explainable Machine Learning Framework for Predicting Auto Loan Defaults. Risks 2025;13:172. https://doi.org/10.3390/risks13090172.
20. Ayari H, Guetari PrR, Kraïem PrN. Machine learning powered financial credit scoring: a systematic literature review. Artif Intell Rev 2025;59:13. https://doi.org/10.1007/s10462-025-11416-2.
21. Mienye ID, Esenogho E, Modisane C. Deep Learning for Credit Risk Prediction: A Survey of Methods, Applications, and Challenges. Information 2026;17:395. https://doi.org/10.3390/info17040395.
22. Hu X, Xu T, Chen Y, Wang S. Predicting Mortgage Default with Machine Learning: AutoML, Class Imbalance, and Leakage Control 2026. https://doi.org/10.48550/ARXIV.2602.00120.
23. Zhang Z, Zheng M, Zhang T, Lin L, Wang Y, Lin L. Bankruptcy Prediction from 10-K Narratives: Evidence from Interpretable Text Scores and Accounting Baselines 2026. https://doi.org/10.48550/ARXIV.2606.05623.
24. Zeng Q, Zhao Z, Yang Y, Zhu Y, Liu F, Bi Z, et al. When AI Meets Wall Street: A Survey on Trustworthy AI in Fintech 2026. https://doi.org/10.48550/ARXIV.2605.30650.
25. Xia Y, Han Z, Li Y, He L. Credit scoring model for fintech lending: An integration of large language models and FocalPoly loss. International Journal of Forecasting 2025;41:894–919. https://doi.org/10.1016/j.ijforecast.2024.07.005.
26. Zeydan İ. Integrating Artificial Intelligence into Business. Journal of Artificial Intelligence and Data Science (JAIDA) 2025;5(2):63-74
27. Tiryaki ŞC, Kavak A. Artificial Intelligence Based Customer Risk Classification for Receivables Management of Businesses. Journal of Artificial Intelligence and Data Science (JAIDA) 2024;4:97–103.
28. Saban S, Dağdevir E. Consumer Preference Prediction with mRMR-Based Explainable EEG Classification. Journal of Artificial Intelligence and Data Science 2025;5(2):125–131.
29. Zakariya G, Akylbekov O, Moldagulova A, Satybaldiyeva R. Innovative Credit Scoring and Sales Accounting Solutions for SMEs in Kazakhstan. FinTech 2025;5:1. https://doi.org/10.3390/fintech5010001.
30. DataSciencePM. What is CRISP DM? DataScience PM 2024. https://www.datascience-pm.com/crisp-dm-2/ (accessed March 8, 2026).
31. Shearer C. The CRISP-DM Model: The New Blueprint for Data Mining. Journal of Data Warehousing 2000;5:13–22.
32. Breiman L, editor. Classification and regression trees. 1. CRC Press repr. Boca Raton, Fla.: Chapman & Hall/CRC; 1984.
33. Breiman L. Bagging predictors. Mach Learn 1996;24:123–40. https://doi.org/10.1007/BF00058655.
34. Liaw A, Wiener M. Package ‘randomForest’ 2022.
35. Quinlan JR. C4.5: Programs for machine learning-. San Mateo, CA, USA: Morgan Kaufman Publishers; 1993.
36. Quinlan JR. Learning decision tree classifiers. ACM Comput Surv 1996;28:71–2. https://doi.org/10.1145/234313.234346.
37. Quinlan R. Data Mining Tools See5 and C5.0 2002.
38. Friedman JH. Greedy function approximation: A gradient boosting machine. Ann Statist 2001;29. https://doi.org/10.1214/aos/1013203451.
39. Breiman L. Random Forests. Machine Learning 2001;45:5–32. https://doi.org/https://doi.org/10.1023/A:1010933404324.
40. Chen T, Guestrin C. XGBoost: A Scalable Tree Boosting System. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Francisco California USA: ACM; 2016, p. 785–94. https://doi.org/10.1145/2939672.2939785.
41. Ke G, Meng Q, Finley T, Wang T, Chen W, Ma W, et al. LightGBM: A Highly Efficient Gradient Boosting Decision Tree. Advances in Neural Information Processing Systems, vol. 30, Https://Papers.Nips.Cc/: NeurIPS 2017; 2017, p. 3146–54.
42. Yandex. CatBoost is a high-performance open source library for gradient boosting on decision trees. CatBoost 2026. https://catboost.ai/ (accessed March 20, 2026).
43. Prokhorenkova L, Gusev G, Vorobev A, Dorogush AV, Gulin A. CatBoost: unbiased boosting with categorical features. Advances in Neural Information Processing Systems 2018;31:6638–48.
44. Hancock JT, Khoshgoftaar TM. CatBoost for big data: an interdisciplinary review. J Big Data 2020;7:94. https://doi.org/10.1186/s40537-020-00369-8.
45. Gafsi N. Machine Learning Approaches to Credit Risk: Comparative Evidence from Participation and Conventional Banks in the UK. JRFM 2025;18:345. https://doi.org/10.3390/jrfm18070345.
46. Cortes C, Vapnik V. Support-vector networks. Mach Learn 1995;20:273–97. https://doi.org/10.1007/BF00994018.
47. Schölkopf B, Smola AJ. Learning with kernels: support vector machines, regularization, optimization, and beyond. Cambridge, Mass: MIT Press; 2002.
48. Schölkopf B, Smola A. Support Vector Machines and Kernel Algorithms 2002.
49. Kononenko I, Kukar M. Chapter 4 - Knowledge Representation. In: Kononenko I, Kukar M, editors. Machine Learning and Data Mining, Woodhead Publishing; 2007, p. 107–29. https://doi.org/https://doi.org/10.1533/9780857099440.107.
50. Domingos P, Pazzani M. On the Optimality of the Simple Bayesian Classifier under Zero-One Loss. Machine Learning 1997;29:103–30. https://doi.org/10.1023/A:1007413511361.
51. Cover T, Hart P. Nearest neighbor pattern classification. IEEE Trans Inform Theory 1967;13:21–7. https://doi.org/10.1109/TIT.1967.1053964.
52. Demirkol D. Diyabetik Ayak Hastalarının Ampütasyon Riskinin Yapay Zekâ Teknikleri ile Öngörülmesi. Doctoral Thesis. İstanbul Üniversitesi, Fen Bilimleri Enstitüsü, Enformatik Anabilim Dalı, 2022.
53. Demirkol D, Erol ÇS, Tannier X, Özcan T, Aktaş Ş. Prediction of amputation risk of patients with diabetic foot using classification algorithms: A clinical study from a tertiary center. International Wound Journal 2024;21:e14556. https://doi.org/10.1111/iwj.14556.
54. Karaboga D. AN IDEA BASED ON HONEY BEE SWARM FOR NUMERICAL OPTIMIZATION. Erciyes University, Engineering Faculty: Erciyes University; 2005.
55. Karaboga D, Gorkemli B, Ozturk C, Karaboga N. A comprehensive survey: artificial bee colony (ABC) algorithm and applications. Artif Intell Rev 2014;42:21–57. https://doi.org/10.1007/s10462-012-9328-0.
56. Karaboga D, Akay B. A comparative study of Artificial Bee Colony algorithm. Applied Mathematics and Computation 2009;214:108–32. https://doi.org/10.1016/j.amc.2009.03.090.
57. Karaboga D, Basturk B. A powerful and efficient algorithm for numerical function optimization: artificial bee colony (ABC) algorithm. J Glob Optim 2007;39:459–71. https://doi.org/10.1007/s10898-007-9149-x.
58. Karaboga D, Basturk B. On the performance of artificial bee colony (ABC) algorithm. Applied Soft Computing 2008;8:687–97. https://doi.org/10.1016/j.asoc.2007.05.007.
59. Konyalıoğlu AK, Apaydın TB, Turhan \.Ilhan, Soydal A, Özcan T. An Extreme Gradient Boosting Model Optimized with Genetic Algorithm for Sales Forecasting of Retail Stores. In: Durakbasa NM, Gençyılmaz MG, editors. Industrial Engineering in the Industry 4.0 Era, Cham: Springer Nature Switzerland; 2024, p. 59–67.
60. Özpolat Z, Yıldırım Ö, Karabatak M. The Effect of Linear Discriminant Analysis and Quantum Feature Maps on QSVM Performance for Obesity Diagnosis. Balkan Journal of Electrical and Computer Engineering 2024;12:206–13. https://doi.org/10.17694/bajece.1475896.
61. Qolomany B, Maabreh M, Al-Fuqaha A, Gupta A, Benhaddou D. Parameters optimization of deep learning models using Particle swarm optimization. 2017 13th International Wireless Communications and Mobile Computing Conference (IWCMC), Valencia, Spain: IEEE; 2017, p. 1285–90. https://doi.org/10.1109/IWCMC.2017.7986470.
62. Rao H, Shi X, Rodrigue AK, Feng J, Xia Y, Elhoseny M, et al. Feature selection based on artificial bee colony and gradient boosting decision tree. Applied Soft Computing 2019;74:634–42. https://doi.org/10.1016/j.asoc.2018.10.036.
63. Mirjalili S, Mirjalili SM, Lewis A. Grey Wolf Optimizer. Advances in Engineering Software 2014;69:46–61. https://doi.org/10.1016/j.advengsoft.2013.12.007.
64. Nematzadeh S, Kiani F, Torkamanian-Afshar M, Aydin N. Tuning hyperparameters of machine learning algorithms and deep neural networks using metaheuristics: A bioinformatics study on biomedical and biological cases. Computational Biology and Chemistry 2022;97:107619. https://doi.org/10.1016/j.compbiolchem.2021.107619.
65. Snoek J, Larochelle H, Adams RP. Practical Bayesian Optimization of Machine Learning Algorithms. Advances in Neural Information Processing Systems, vol. 25, 2012.
66. Shahriari B, Swersky K, Wang Z, Adams RP, De Freitas N. Taking the Human Out of the Loop: A Review of Bayesian Optimization. Proc IEEE 2016;104:148–75. https://doi.org/10.1109/JPROC.2015.2494218.
67. Frazier PI. A Tutorial on Bayesian Optimization 2018. https://doi.org/https://doi.org/10.48550/arXiv.1807.02811.
68. Kennedy J, Eberhart R. Particle swarm optimization. Proceedings of ICNN’95 - International Conference on Neural Networks, vol. 4, 1995, p. 1942–8 vol.4. https://doi.org/10.1109/ICNN.1995.488968.
69. Karaboğa D. Yapay Zekâ Optimizasyon Algoritmaları. 6. Baskı. NOBEL AKADEMİK YAYINCILIK; 2018.
70. Lundberg SM, Erion G, Chen H, DeGrave A, Prutkin JM, Nair B, et al. From local explanations to global understanding with explainable AI for trees. Nat Mach Intell 2020;2:56–67. https://doi.org/10.1038/s42256-019-0138-9.
71. Lundberg SM, Lee S-I. A Unified Approach to Interpreting Model Predictions. Advances in Neural Information Processing Systems 30 (NeurIPS 2017), Https://Neurips.Cc/Conferences/2017: NeurIPS 2017; 2017.