Evaluación Inteligente del Riesgo Crediticio: Un Enfoque Explicable con Aprendizaje Automático y Modelos de Lenguaje de Gran Tamaño

Henry Guamán; Marcos Orellana

doi:10.56048/MQR20225.9.3.2025.e877

Autores/as

Henry Guamán UNIVERSIDAD INDOAMÉRICA https://orcid.org/0009-0007-0608-7425
Marcos Orellana UNIVERSIDAD DEL AZUAY https://orcid.org/0000-0002-3671-9362

DOI:

https://doi.org/10.56048/MQR20225.9.3.2025.e877

Palabras clave:

Aprendizaje automático; Árbol de decisión; Cooperativas; Evaluación crediticia; Inteligencia artificial; Riesgo crediticio

Resumen

Este estudio aborda la evaluación del riesgo crediticio en cooperativas mediante la integración de un modelo de árbol de decisión explicable y un chatbot basado en Phi4. Se utilizó un conjunto de datos de 5000 registros, considerando variables clave como edad, número de dependientes, antigüedad en la institución, monto solicitado, segmento crediticio y duración del empleo. El modelo de árbol de decisión, configurado con parámetros optimizados (profundidad máxima, ccp_alpha, número mínimo de muestras para división y hojas), alcanzó una precisión del 90,47%, demostrando su capacidad para discriminar entre clientes “seguros” e “inseguros”. Además, se evaluaron otros modelos (OneR, PART y PRISM), siendo PART y el árbol de decisión los que presentaron el mejor equilibrio entre precisión e interpretabilidad. La incorporación del chatbot, entrenado mediante técnicas de transferencia de aprendizaje y desplegado en un entorno local seguro, proporcionó explicaciones claras sobre las decisiones crediticias, facilitando auditorías por organismos reguladores. La propuesta destaca la importancia de emplear enfoques de inteligencia artificial explicable (XAI) para mejorar la inclusión financiera, optimizar recursos y reducir los tiempos de procesamiento. Se reconocen limitaciones relacionadas con la calidad del conjunto de datos y se sugiere integrar modelos híbridos y expandir las fuentes de información en futuras investigaciones para lograr evaluaciones más robustas y adaptables. Este enfoque integral mejora significativamente la eficiencia y la transparencia en la gestión.

Descargas

Los datos de descargas todavía no están disponibles.

Métricas

Cargando métricas ...

Cited

DOI: 10.56048

Biografía del autor/a

Henry Guamán, UNIVERSIDAD INDOAMÉRICA

Facultad de Ingenierías

Marcos Orellana, UNIVERSIDAD DEL AZUAY

Computer Science Research & Development Laboratory (LIDI),

Citas

Abdin, M., Aneja, J., & Behl, H., et al. (2024). Phi-4 Technical Report: A 14-billion parameter language model developed with a training recipe focused on data quality. arXiv preprint arXiv:2412.08905. https://doi.org/10.48550/arXiv.2412.08905

Al Khaldy, M., et al. (2024). Securing digital finance: Applying machine learning for fraud analysis. In 2nd International Conference on Cyber Resilience (ICCR). https://doi.org/10.1109/ICCR61006.2024.10533140

Al-Ameer, A., & Al-Sunni, F. (2021). A methodology for securities and cryptocurrency trading using exploratory data analysis and artificial intelligence. In 2021 1st International Conference on Artificial Intelligence and Data Analytics (CAIDA) (pp. 54–61). https://doi.org/10.1109/CAIDA51941.2021.9425223

Anusha, H., & Bhowmik, B. (2023). Feature selection for peer-to-peer lending default risk using Boruta and mRMR approach. In 2023 IEEE 20th India Council International Conference (INDICON) (pp. 983–988). https://doi.org/10.1109/INDICON59947.2023.10440917

Chaudhary, D., & Saroj, S. K. (2023). Cryptocurrency price prediction using supervised machine learning algorithms. Advances in Distributed Computing and Artificial Intelligence Journal, 12, 31490. https://doi.org/10.14201/adcaij.31490

Chen, H.-Y., et al. (2024). Skewness-aware boosting regression trees for customer contribution prediction in financial precision marketing. In WWW 2024 Companion – Proceedings of the ACM Web Conference (pp. 461–470). https://doi.org/10.1145/3589335.3648346

Edo, O. C., et al. (2023). Fintech adoption dynamics in a pandemic: An experience from some financial institutions in Nigeria during COVID-19 using machine learning approach. Cogent Business and Management, 10, 2242985. https://doi.org/10.1080/23311975.2023.2242985

IC. (2023). 6th International Conference on Innovative Computing. Lecture Notes in Electrical Engineering, 1044, 90–98.

IC. (2023). 6th International Conference on Innovative Computing. Lecture Notes in Electrical Engineering, 1045, 100–110.

ICACNIS. (2022). Blockchain technology, intelligent systems, and the applications for human life. In Proceedings of ICACNIS 2022 – International Conference on Advanced Creative Networks and Intelligent Systems (pp. 50–60).

ICETSIS. (2024). 2024 ASU International Conference in Emerging Technologies for Sustainability and Intelligent Systems. In Proceedings of ICETSIS 2024 (pp. 10–15).

Karn, A. L., et al. (2022). Designing a deep learning-based financial decision support system for fintech to support corporate customer’s credit extension. Malaysian Journal of Computer Science, 1(Special Issue 1), 116–131. https://doi.org/10.22452/mjcs.sp2022no1.9

Ke, L., et al. (2021). Loan repayment behavior prediction of provident fund users using a stacking-based model. In 2021 IEEE 6th International Conference on Cloud Computing and Big Data Analytics (ICCCBDA) (pp. 37–43). https://doi.org/10.1109/ICCCBDA51879.2021.9442613

Khan, W., et al. (2022). Stock market prediction using machine learning classifiers and social media, news. Journal of Ambient Intelligence and Humanized Computing, 13, 3433–3456. https://doi.org/10.1007/s12652-020-01839-w

Khenfouci, Y., & Challal, Y. (2023). SuperChain: Decentralized and trustful supervised learning over blockchain. In 2023 5th International Conference on Blockchain Computing and Applications (BCCA) (pp. 627–634). https://doi.org/10.1109/BCCA58897.2023.10338875

Li, L. (2023). Investigating the application of 3D avatar chatbot services on fintech. In Proceedings of the 29th Annual Americas Conference on Information Systems (AMCIS) (pp. 45–52).

Liu, J., et al. (2023). Interpreting the prediction results of the tree-based gradient boosting models for financial distress prediction with an explainable machine learning approach. Journal of Forecasting, 42, 1112–1137. https://doi.org/10.1002/for.2931

Liu, Y. (2024). Discussion on the enterprise financial risk management framework based on AI fintech. Decision Making: Applications in Management and Engineering, 7, 254–269. https://doi.org/10.31181/dmame712024942

Liu, Y., et al. (2021). A deep neural network based model for stock market prediction. In 2021 IEEE 2nd International Conference on Big Data, Artificial Intelligence and Internet of Things Engineering (ICBAIE) (pp. 320–323). https://doi.org/10.1109/ICBAIE52039.2021.9390010

Lusinga, M., et al. (2021). Investigating statistical and machine learning techniques to improve the credit approval process in developing countries. In IEEE AFRICON Conference (pp. 200–207). https://doi.org/10.1109/AFRICON51333.2021.9570906

Majumder, A., et al. (2022). Stock market prediction: A time series analysis. Smart Innovation, Systems and Technologies, 235, 389–401. https://doi.org/10.1007/978-981-16-2877-1_35

Malakauskas, A., & Lakstutiene, A. (2021). The application of artificial intelligence tools in creditworthiness modelling for SME entities. In IEEE International Conference on Technology and Entrepreneurship (ICTE 2021) (pp. 120–126). https://doi.org/10.1109/ICTE51655.2021.9584528

Paul, M., et al. (2024). Trustworthy deep learning techniques for credit risk assessment. In 5th International Conference on Electronics and Sustainable Communication Systems (ICESC) (pp. 1949–1962). https://doi.org/10.1109/ICESC60852.2024.10689741

Rabbani, H., et al. (2024). Enhancing security in financial transactions: A novel blockchain-based federated learning framework for detecting counterfeit data in fintech. PeerJ Computer Science, 10, e2280. https://doi.org/10.7717/peerj-cs.2280

Shetu, S. F., et al. (2021). Predicting satisfaction of online banking system in Bangladesh by machine learning. In ICAICST 2021 – International Conference on Artificial Intelligence and Computer Science Technology (pp. 223–228). https://doi.org/10.1109/ICAICST53116.2021.9497796

Shen, Y. (2022). Application of supplemental sampling and interpretable AI in credit scoring for Canadian fintechs: Methods and case studies. In Lecture Notes in Computer Science, 13725 (pp. 3–14). https://doi.org/10.1007/978-3-031-22064-7_1

Silva, C., et al. (2023). Towards interpretability in fintech applications via knowledge augmentation. In Lecture Notes in Computer Science (LNCS), 14115 (pp. 106–117). https://doi.org/10.1007/978-3-031-49008-8_9

Tahyudin, I., et al. (2023). Sentiment analysis model development on E-Money service complaints. TEM Journal, 12, 2050–2055. https://doi.org/10.18421/TEM124-15

Xia, Y., et al. (2021). Incorporating multilevel macroeconomic variables into credit scoring for online consumer lending. Electronic Commerce Research and Applications, 49, 101095. https://doi.org/10.1016/j.elerap.2021.101095

Zhao, X., & Zhao, Q. (2021). Stock prediction using optimized LightGBM based on cost awareness. In 2021 5th IEEE International Conference on Cybernetics (CYBCONF) (pp. 107–113). https://doi.org/10.1109/CYBCONF51991.2021.9464148

Zhu, Y., & Tingting, N. (2024). Application of back propagation neural network method in digital economy prediction. In 3rd IEEE International Conference on Distributed Computing and Electrical Circuits and Electronics (ICDCECE) (pp. 1–8). https://doi.org/10.1109/ICDCECE60827.2024.10548494

Evaluación Inteligente del Riesgo Crediticio: Un Enfoque Explicable con Aprendizaje Automático y Modelos de Lenguaje de Gran Tamaño

Autores/as

DOI:

Palabras clave: