Beyond transaction borders: Tabular and graph-based machine learning for multi-domain financial crime detection
DOI:
https://doi.org/10.33003/fujafr-2026.v4i3.407.138-145Keywords:
anti-money laundering, fraud detection, graph neural networks, GraphSAGE, class imbalance, Elliptic Bitcoin datasetAbstract
Purpose: This study examines how reported detection performance in financial crime machine learning depends on the evaluation protocol and assesses when relational structure adds detection value.
Methodology: Three public benchmarks are evaluated: Credit Card Fraud 2023, the Synthetic Anti-Money Laundering Dataset (SAML-D), and the Elliptic Bitcoin Dataset. Tabular models (XGBoost, LightGBM, Random Forest) are compared under SMOTE, class weighting and threshold calibration. Graph models (GCN, GraphSAGE) run on samples matched to the tabular ones in size and class prevalence. Every experiment uses a three-way train, validation and test partition across five random seeds, derives all feature statistics from the training fold, and reports average precision. Random and chronological splits are run side by side. A no-edge ablation isolates the contribution of graph structure.
Results and conclusion: Threshold calibration yields no measurable gain for XGBoost on Credit Card Fraud and raises Random Forest F1 there from 0.8470 to 0.8748. On SAML-D under chronological evaluation it raises XGBoost F1 from 0.1380 to 0.4717, a factor of 3.4. Class separability appears to help explain this difference. Graph structure contributes a gain of 0.46 in average precision on SAML-D under chronological evaluation, with no clear gain on Elliptic.
Implication of findings: Detection figures obtained under random splits may overstate prospective performance. Compliance functions assessing internal or vendor models are well served by requiring chronological validation, threshold selection on held-out data, and precision-recall reporting before accepting performance claims
References
Abubakar, A., Abdullahi, A., Suleiman, H., & Abdullahi, M. (2026). Basel III regulatory reforms and bank financial performance: Evidence from Nigerian deposit money banks. FUDMA Journal of Accounting and Finance Research, 4(2), 92-105. https://doi.org/10.33003/fujafr-2026.v4i2.369.92-105
Akpootu, E. J., & Yusuf, Y. (2025). Forensic accounting litigation support and detection of public sector corruption in Nigeria. FUDMA Journal of Accounting and Finance Research, 3(4), 221-238. https://doi.org/10.33003/fujafr-2025.v3i4.253.221-238
Alarab, I., Prakoonwit, S., & Nacer, M. I. (2020). Competence of graph convolutional networks for anti-money laundering in Bitcoin blockchain. Proceedings of the 2020 5th International Conference on Machine Learning Technologies, 23-27. https://doi.org/10.1145/3409073.3409080
Bergmeir, C., & Benitez, J. M. (2012). On the use of cross-validation for time series predictor evaluation. Information Sciences, 191, 192-213. https://doi.org/10.1016/j.ins.2011.12.028
Board of Governors of the Federal Reserve System & Office of the Comptroller of the Currency. (2011). Supervisory guidance on model risk management (SR 11-7; OCC Bulletin 2011-12). Superseded by SR 26-2 (2026).
Board of Governors of the Federal Reserve System, Office of the Comptroller of the Currency, & Federal Deposit Insurance Corporation. (2026). Revised guidance on model risk management (SR 26-2). https://www.federalreserve.gov/supervisionreg/srletters/SR2602.htm
Branco, P., Torgo, L., & Ribeiro, R. P. (2016). A survey of predictive modeling on imbalanced domains. ACM Computing Surveys, 49(2), Article 31. https://doi.org/10.1145/2907070
Breiman, L. (2001). Random forests. Machine Learning, 45(1), 5-32. https://doi.org/10.1023/A:1010933404324
Chawla, N. V., Bowyer, K. W., Hall, L. O., & Kegelmeyer, W. P. (2002). SMOTE: Synthetic minority over-sampling technique. Journal of Artificial Intelligence Research, 16, 321-357. https://doi.org/10.1613/jair.953
Chen, T., & Guestrin, C. (2016). XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785-794. https://doi.org/10.1145/2939672.2939785
Dal Pozzolo, A., Boracchi, G., Caelen, O., Alippi, C., & Bontempi, G. (2018). Credit card fraud detection: A realistic modeling and a novel learning strategy. IEEE Transactions on Neural Networks and Learning Systems, 29(8), 3784-3797. https://doi.org/10.1109/TNNLS.2017.2736643
Elgiriyewithana, N. (2023). Credit card fraud detection dataset 2023 [Data set]. Kaggle. https://doi.org/10.34740/KAGGLE/DSV/6492730
Elmougy, Y., & Liu, L. (2023). Demystifying fraudulent transactions and illicit nodes in the Bitcoin network for financial forensics. Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 3979-3990. https://doi.org/10.1145/3580305.3599803
Financial Action Task Force. (2021). Opportunities and challenges of new technologies for AML/CFT. FATF.
Gita, E. E., & Oriavwote, V. (2025). Digital banking tools and savings performance. FUDMA Journal of Accounting and Finance Research, 3(4), 35-44. https://doi.org/10.33003/fujafr-2025.v3i4.235.35-44
Hamilton, W. L., Ying, R., & Leskovec, J. (2017). Inductive representation learning on large graphs. Advances in Neural Information Processing Systems, 30, 1024-1034.
Kaufman, S., Rosset, S., Perlich, C., & Stitelman, O. (2012). Leakage in data mining: Formulation, detection, and avoidance. ACM Transactions on Knowledge Discovery from Data, 6(4), Article 15. https://doi.org/10.1145/2382577.2382579
Kenah, G. H., & Mohe, E. (2025). Effect of responsibility accounting on financial and social performances of branch expansion of microfinance institutions in the CamCCUL network, Cameroon. FUDMA Journal of Accounting and Finance Research, 3(4), 293-310. https://doi.org/10.33003/fujafr-2025.v3i4.264.293-310
Kipf, T. N., & Welling, M. (2017). Semi-supervised classification with graph convolutional networks. International Conference on Learning Representations.
Le, H. S., & Oni, D. (2026). Multimedia multimodal artificial intelligence (MMAI): Foundations, challenges, and future directions. Journal of Computer Science and Cybernetics, 42(1), 1-24. https://doi.org/10.15625/1813-9663/23021
Lo, W. W., Kulatilleke, G. K., Sarhan, M., Layeghy, S., & Portmann, M. (2023). Inspection-L: Self-supervised GNN node embeddings for money laundering detection in Bitcoin. Applied Intelligence, 53(16), 19406-19417. https://doi.org/10.1007/s10489-023-04504-9
Lundberg, S. M., & Lee, S.-I. (2017). A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 30, 4765-4774.
Nilson Report. (2023, December). Card fraud losses worldwide. Issue 1254.
Oni, D., Arshad, E., & Pham, B. N. (2023). Cybercrime on social media in Nigeria: Trends, scams, vulnerabilities and prevention. Proceedings of the Cyber Secure Nigeria Conference, 143-150. https://doi.org/10.22624/AIMS/CSEAN-SMART2023P17
Oztas, B., Cetinkaya, D., Adedoyin, F., Budka, M., Aksu, G., & Dogan, H. (2024). Transaction monitoring in anti-money laundering: A qualitative analysis and points of view from industry. Future Generation Computer Systems, 159, 161-171. https://doi.org/10.1016/j.future.2024.05.027
Oztas, B., Cetinkaya, D., Adedoyin, F., Budka, M., Dogan, H., & Aksu, G. (2023). Enhancing anti-money laundering: Development of a synthetic transaction monitoring dataset. 2023 IEEE International Conference on e-Business Engineering (ICEBE), 47-54. https://doi.org/10.1109/ICEBE59045.2023.00028
Oztas, B., Cetinkaya, D., Adedoyin, F., Budka, M., Dogan, H., & Aksu, G. (2025). Tab-AML: A transformer based transaction monitoring model for anti-money laundering. 2025 IEEE Conference on Artificial Intelligence (CAI), 161-167. https://doi.org/10.1109/CAI64502.2025.00033
Saito, T., & Rehmsmeier, M. (2015). The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets. PLOS ONE, 10(3), e0118432. https://doi.org/10.1371/journal.pone.0118432
United Nations Office on Drugs and Crime. (2023). Money laundering overview. https://www.unodc.org/unodc/en/money-laundering/overview.html
Weber, M., Domeniconi, G., Chen, J., Weidele, D. K. I., Bellei, C., Robinson, T., & Leiserson, C. E. (2019). Anti-money laundering in Bitcoin: Experimenting with graph convolutional networks for financial forensics. KDD Workshop on Anomaly Detection in Finance.
Zhou, J., Cui, G., Hu, S., Zhang, Z., Yang, C., Liu, Z., Wang, L., Li, C., & Sun, M. (2020). Graph neural networks: A review of methods and applications. AI Open, 1, 57-81. https://doi.org/10.1016/j.aiopen.2021.01.001
Published
Issue
Section
License
Copyright (c) 2026 Olubunmi Grace Oni, Aderemi Rasheedat Toyin, Damilola I. Oni
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
The FUDMA Journal of Accounting and Finance Research (FUJAFR) operates a copyright policy that ensures a balance between author rights and wide dissemination of scholarly work.
1. Author Copyright Retention
Authors retain full copyright of their published work without restriction. Submission to FUJAFR does not transfer ownership of copyright to the journal.
2. License to Publish
By submitting a manuscript and upon acceptance, authors grant FUJAFR:
- The right to publish, reproduce, and distribute the article
- The right to identify itself as the original publisher of the work
This grant is non-exclusive, meaning authors are free to reuse their work in other contexts, provided proper acknowledgment of the original publication in FUJAFR is made.
3. Licensing of Published Content
All articles are published under the:
Creative Commons Attribution-NonCommercial License (CC BY-NC 4.0)
Under this license:
- Users may share and adapt the work for non-commercial purposes only
- Proper attribution to the author(s) and the journal is required
- Any commercial use requires explicit permission from the copyright holder
4. Author Reuse Rights
Authors are permitted to:
- Archive their published articles in institutional repositories or personal websites
- Share their work for educational and research purposes
- Reuse portions of their work in future publications (e.g., books or other articles), provided proper citation of the original publication is included
5. Third-Party Content
Authors are responsible for obtaining permission to use any third-party copyrighted material (e.g., images, tables, datasets) included in their manuscripts. Proper acknowledgment must be provided where required.
6. Attribution Requirement
All users of FUJAFR content must provide appropriate credit, including:
- Author name(s)
- Article title
- Journal name (FUJAFR)
- Year of publication
