A Study on Credit Default Prediction Using Hybrid AI Models Combining Neural Architectures and Econometric Features
DOI:
https://doi.org/10.63282/3050-922X.IJERET-V6I2P110Keywords:
Credit Default Prediction, Hybrid AI Models, Econometric Features, Deep Learning, Neural Networks, Logistic Regression, LSTM, Financial Risk Modeling, AUC-ROC, Interpretable AIAbstract
Credit default prediction plays a vital role in risk management, lending strategies, and financial stability. While traditional econometric models offer interpretability, they often lack the predictive power of contemporary neural networks. This study proposes a novel hybrid approach that integrates deep neural network (DNN) architectures with key econometric indicators to improve the prediction of credit default risk. We develop and evaluate several model configurations, including a hybrid Long Short-Term Memory (LSTM) and logistic regression framework, on publicly available credit datasets. The results show that hybrid models outperform standalone econometric or deep learning models in terms of accuracy, AUC-ROC, and F1-score. The study also explores feature importance to enhance model explainability. Our findings underscore the potential of combining statistical and AI methodologies for more accurate and interpretable financial risk assessments
References
[1] Altman, E. (1968). Financial ratios, discriminant analysis and the prediction of corporate bankruptcy. The Journal of Finance, 23(4), 589–609.
[2] Thomas, L. C., Edelman, D. B., & Crook, J. N. (2002). Credit Scoring and Its Applications. SIAM.
[3] Tian, S., Yu, Y., & Gu, D. (2015). Variable selection and corporate bankruptcy forecasts. Computational Economics, 45(1), 41–62.
[4] Yeh, I. C., & Lien, C. H. (2009). The comparisons of data mining techniques for the predictive accuracy of probability of default of credit card clients. Expert Systems with Applications, 36(2), 2473–2480.
[5] Lessmann, S., Baesens, B., Seow, H. V., & Thomas, L. C. (2015). Benchmarking state-of-the-art classification algorithms for credit scoring. European Journal of Operational Research, 247(1), 124–136.
[6] Brown, I., & Mues, C. (2012). An experimental comparison of classification algorithms for imbalanced credit scoring data sets. Expert Systems with Applications, 39(3), 3446–3453.
[7] Kumar, A.; Sharma, S.; Mahdavi, M. Machine Learning (ML) Technologies for Digital Credit Scoring in Rural Finance: A Literature Review. Risks 2021, 9, 192. https://doi.org/10.3390/risks9110192
[8] Longyue Liang. et al. 2020. Forecasting peer-to-peer platform default rate with LSTM neural network. Electronic Commerce Research and Applications. Volume 43, September–October 2020, 100997. https://doi.org/10.1016/j.elerap.2020.100997
[9] Ala’raj, M., & Abbod, M. (2016). A new hybrid ensemble credit scoring model based on classifiers consensus system approach. Expert Systems with Applications, 64, 36–55.
[10] Bo-Wen Chi. et la. 2012. A hybrid approach to integrate genetic algorithm into dual scoring model in enhancing the performance of credit scoring model. Expert Systems with Applications. Volume 39, Issue 3, 15 February 2012, Pages 2650-2661. https://doi.org/10.1016/j.eswa.2011.08.120
[11] Bo-Wen Chi and Chiun-Chieh Hsu. 2012. A hybrid approach to integrate genetic algorithm into dual scoring model in enhancing the performance of credit scoring model. Expert Syst. Appl. 39, 3 (February, 2012), 2650–2661. https://doi.org/10.1016/j.eswa.2011.08.120
[12] https://bura.brunel.ac.uk/bitstream/2438/17809/2/FullText.pdf
