Impact of Synthetic Data Generated Using CTGAN on Machine Learning Models for Prediction
DOI:
https://doi.org/10.35381/i.p.v8i15.5249Keywords:
Artificial intelligence, data processing, data analysis, statistics, (UNESCO Thesaurus)Abstract
The objective of this study was to evaluate comparatively the predictive performance and explainability of Logistic Regression, Random Forest, and XGBoost trained on real data versus real data augmented with CTGAN. This research is computational/experimental in nature, focusing on predictive models, with a quantitative approach, a comparative purpose, and an applied orientation. The results revealed findings contrary to what is reported in most of the literature reviewed: adding synthetic data to the training did not improve the predictive performance of any of the three models but rather reduced it in a statistically significant manner, even when that same synthetic data passed statistical fidelity validation compared to the real data. In conclusion, at the theoretical level, the study provides a concrete example of the relationship between statistical fidelity and the predictive utility of synthetic data an aspect that remains largely unexplored in the reviewed literature.
Downloads
References
Adiputra, I. N. M., Lin, P.-C., & Wanchai, P. (2025). The effectiveness of generative adversarial network-based oversampling methods for imbalanced multi-class credit score classification. Electronics, 14(4), Article 697. https://doi.org/10.3390/electronics14040697
Ali Shahee, S., & Patel, R. (2025). An explainable ADASYN-based focal loss approach for credit assessment. Journal of Forecasting, 44(4), 1513–1530. https://doi.org/10.1002/for.3252
Borisov, V., Leemann, T., Seßler, K., Haug, J., Pawelczyk, M., & Kasneci, G. (2024). Deep neural networks and tabular data: A survey. IEEE Transactions on Neural Networks and Learning Systems, 35(6), 7499–7519. https://n9.cl/khip9
Breiman, L. (2001). Random forests. Machine Learning, 45(1), 5–32. https://doi.org/10.1023/A:1010933404324
Chen, T., & Guestrin, C. (2016). XGBoost: A scalable tree boosting system. En Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (pp. 785–794). https://doi.org/10.1145/2939672.2939785
El Sayad, I., Chong, F. H. L., & Gosal, B. (2026). Risk assessment in mortgages: A comparative study of AI models. PeerJ Computer Science, 12, Article e3494. https://doi.org/10.7717/peerj-cs.3494
Gao, X., & Zhou, Y. (2026). A resampling ensemble model for multi-window corporate default prediction under class imbalance. Systems, 14(7), Article 776. https://n9.cl/nk6n55
Gómez-de-Mariscal, E., Guerrero, V., Sneider, A., Jayatilaka, H., Phillip, J. M., Wirtz, D., & Muñoz-Barrutia, A. (2021). Use of the p-values as a size-dependent function to address practical differences when analyzing large datasets. Scientific Reports, 11, Article 20942. https://n9.cl/je6yl
Kheddar, H., Himeur, Y., Al-Maadeed, S., Amira, A., & Bensaali, F. (2023). Deep transfer learning for automatic speech recognition: Towards better generalization. Knowledge-Based Systems, 277, Article 110851. https://doi.org/10.1016/j.knosys.2023.110851
Li, Z., Chen, G., Liu, Y., & Yuan, J. (2025). An efficient multi-institution collaborative learning framework based on knowledge filtering and data sharing. Cluster Computing, 28(10), Article 644. https://n9.cl/wmjvdd
Lundberg, S. M., & Lee, S.-I. (2017). A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 30. https://n9.cl/82snlp
Petersen, K., Feldt, R., Mujtaba, S., & Mattsson, M. (2008). Systematic mapping studies in software engineering. En 12th International Conference on Evaluation and Assessment in Software Engineering (EASE). BCS Learning & Development. https://doi.org/10.14236/ewic/EASE2008.8
Petticrew, M., & Roberts, H. (2006). Systematic reviews in the social sciences: A practical guide. Blackwell Publishing. https://n9.cl/rw7s1r
Pourkhoshgoftar, S., Shahbahrami, A., & Esmi, N. (2025). Graph-based inductive learning for credit risk prediction with imbalance mitigation. Computational Economics. https://doi.org/10.1007/s10614-025-11114-9
Raheem, E. A., Dinar, A. M., Mohammed, M. A., & Al-Attar, B. (2024). Improving loan status prediction accuracy with generative adversarial networks: Addressing data scarcity and bias. Journal of Intelligent Systems and Internet of Things, 13(1), 251–258. https://doi.org/10.54216/JISIoT.130118
Ramachandra, P., & Vaithiyanathan, S. (2025). Fed-DPSDG-WGAN: Differentially private synthetic data generation for loan default prediction via federated Wasserstein GAN. IEEE Access, 13, 52069–52084. https://n9.cl/knhnf
Shan, W., Kang, X., & Gao, B. (2026). A TabPFN-based framework for credit risk prediction in automotive green supply chain finance. Sustainability, 18(12), Article 6305. https://doi.org/10.3390/su18126305
Singhal, P., Walambe, R., Ramanna, S., & Kotecha, K. (2023). Domain adaptation: Challenges, methods, datasets, and applications. IEEE Access, 11, 6973–7020. https://n9.cl/3wvi76
van Eck, N. J., & Waltman, L. (2010). Software survey: VOSviewer, a computer program for bibliometric mapping. Scientometrics, 84(2), 523–538. https://doi.org/10.1007/s11192-009-0146-3
Xu, L., Skoularidou, M., Cuesta-Infante, A., & Veeramachaneni, K. (2019). Modeling tabular data using conditional GAN. Advances in Neural Information Processing Systems, 32. https://n9.cl/rfe5on
Xueping, Z., Samuri, S. M., & Adnan, M. H. M. (2026). Adaptive anomaly correction and dynamic sample balancing for deep intelligent credit risk assessment framework. NED University Journal of Research, 23(1), 44–69. https://doi.org/10.35453/NEDJR-ASCN-2025-007.R4
Zhao, X., & Guan, S. (2023). CTCN: A novel credit card fraud detection method based on Conditional Tabular Generative Adversarial Networks and Temporal Convolutional Network. PeerJ Computer Science, 9, Article e1634. https://doi.org/10.7717/peerj-cs.1634
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Wilmer Illescas-Espinoza, Nancy Loja-Mora, Wilmer Rivas-Asanza
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
CC BY-NC-SA : Esta licencia permite a los reutilizadores distribuir, remezclar, adaptar y construir sobre el material en cualquier medio o formato solo con fines no comerciales, y solo siempre y cuando se dé la atribución al creador. Si remezcla, adapta o construye sobre el material, debe licenciar el material modificado bajo términos idénticos.
OAI-PMH URL: https://fundacionkoinonia.com.ve/ojs/index.php/ingeniumetpotentia/oai
