Designing an Intelligent Support System to Identify the Financial Health of Bank Customers Based on Ensemble Deep Learning
Keywords:
Decision Support System, Group Deep Learning, Financial Health, Credit RiskAbstract
Objective: This study aimed to design an intelligent decision support system for identifying the financial health and risk priority of bank customers by integrating the expert knowledge of credit assessment professionals with ensemble deep learning models.
Methodology: This study employed an exploratory sequential mixed-methods design in two qualitative and quantitative phases. In the qualitative phase, data were collected through 16 semi-structured interviews with credit assessment and credit risk management experts working in commercial and specialized banks in Tehran and were analyzed using Graneheim and Lundman’s conventional content analysis method in MAXQDA 2022. In the quantitative phase, credit and transaction data from 4,872 individual customer credit files of a large Iranian commercial bank were extracted. A three-branch hybrid architecture was developed, including a bidirectional LSTM network for modeling 12-month financial behavior trends, a fully connected network for static credit indicators, and another fully connected network for demographic variables. The dataset was divided into training, validation, and test sets, and model performance was assessed using ROC-AUC, precision, recall, F1-Score, confusion matrix, and SHAP-based explainability analysis.
Findings: The proposed hybrid model achieved a mean ROC-AUC of 0.93 and a weighted F1-Score of 0.86 on the test set. The F1-Score values for low-risk, high-risk, and critical customers were 0.93, 0.83, and 0.76, respectively. SHAP analysis indicated that debt-to-income ratio, installment coverage trend, and credit score collectively accounted for approximately 60% of the model’s predictive power. Moreover, the LSTM branch alone improved predictive performance by 0.23 compared with the static baseline model.
Conclusion: Integrating tacit banking expertise with hybrid deep learning architectures can support the development of accurate, explainable, and reliable decision support systems for diagnosing customer financial health and improving credit risk management.
Downloads
References
Alahmari, F., Saad, S., & Pileggi, S. F. (2023). A systematic literature review of tacit knowledge management in the banking sector. Journal of Knowledge Management, 27(5), 1310-1334.
Arrieta, A. B., Diaz-Rodriguez, N., Del Ser, J., Bennetot, A., Tabik, S., Barbado, A., & Herrera, F. (2020). Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI. Information Fusion, 58, 82-115. https://doi.org/10.1016/j.inffus.2019.12.012
Berger, A. N., Molyneux, P., & Wilson, J. O. S. (2020). Banks and the real economy: An assessment of the research. Journal of Corporate Finance, 62, 101513. https://doi.org/10.1016/j.jcorpfin.2019.101513
Dastile, X., & Celik, T. (2021). Making deep learning-based predictions for credit scoring explainable. IEEE Access, 9, 50426-50440. https://doi.org/10.1109/ACCESS.2021.3068854
Dastile, X., Celik, T., & Potsane, M. (2020). Statistical and machine learning models in credit scoring: A systematic literature review. Applied Soft Computing, 91, 106263. https://doi.org/10.1016/j.asoc.2020.106263
Gicic, A., & Subasi, A. (2023). Credit scoring for a microcredit data set using deep learning ensemble methods. Computing in science & engineering, 25(2), 20-31.
Graneheim, U. H., & Lundman, B. (2004). Qualitative content analysis in nursing research: Concepts, procedures and measures to achieve trustworthiness. Nurse Education Today, 24(2), 105-112. https://doi.org/10.1016/j.nedt.2003.10.001
Gunnarsson, B. R., Vanden Broucke, S., Baesens, B., Oskarsdottir, M., & Lemahieu, W. (2021). Deep learning for credit scoring: Do or don't? European Journal of Operational Research, 295(1), 292-305. https://doi.org/10.1016/j.ejor.2021.03.006
Heaton, J. B., Polson, N. G., & Witte, J. H. (2017). Deep learning for finance: Deep portfolios. Applied Stochastic Models in Business and Industry, 33(1), 3-12. https://doi.org/10.1002/asmb.2209
Heaton, J. B., Polson, N. G., & Witte, J. H. (2021). Deep learning in finance: A review. Journal of Financial Data Science, 3(2), 10-25.
Maziriri, E. T., Rukuni, T. F., Nyagadza, B., & Bepe, T. (2026). The customer's quest to manage funds digitally: Exploring the factors that influence the intention to use and the ongoing use of mobile banking apps. Computers in Human Behavior Reports, 21, 100907. https://doi.org/10.1016/j.chbr.2025.100907
Petropoulos, A., Siakoulis, V., Stavroulakis, E., & Vlachogiannakis, N. E. (2020). Predicting bank insolvencies using machine learning techniques. International Journal of Forecasting, 36(3), 1092-1113. https://doi.org/10.1016/j.ijforecast.2019.11.005
Sadhwani, A., Giesecke, K., & Sirignano, J. (2021). Deep learning for mortgage risk. Journal of Financial Econometrics, 19(2), 313-341. https://doi.org/10.1093/jjfinec/nbaa025
Shadzad, V., Naami, A., Kimasi, M., & Azad, N. (2025). Designing a digital content marketing model with a customer engagement approach in commercial banks. Business, Marketing, and Finance Open, 2(1), 150-160. https://doi.org/10.61838/bmfopen.2.1.14
Shafiei, A. S. M., Taleghani, M., & Haghayegh, R. A. S. (2025). Designing and Validating a Behavioral Model of Banking System Customers. Management Strategies and Engineering Sciences, 7(3), 47-56. https://doi.org/10.61838/msesj.7.3.5
Taghipourian, M. J. (2026). The role of emotional branding on brand equity in the banking sector: Is customer experience a mediator? Dynamic Management and Business Analysis, 1-17. https://www.dmbaj.com/index.php/dmba/article/view/247
Yan, X., Li, Y., Nie, F., & Li, R. (2025). Bank Customer Segmentation and Marketing Strategies Based on Improved DBSCAN Algorithm. Applied Sciences, 15(6).
Downloads
Published
Submitted
Revised
Accepted
Issue
Section
License
Copyright (c) 2025 Aliasghar Gharaei, Maryam Rahmaty, Seyed Ahmad Shayannia (Author)
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
