EVALUATION AND DEVELOPMENT OF MACHINE-LEARNING-BASED COMPLIANCE SCORING MODELS FOR ALCOHOLIC BEVERAGES EXCISE TAXPAYER IN INDONESIA WITH INNOVATIVE ENCODING STRATEGY
DOI:
https://doi.org/10.25077/jmua.15.4.495-505.2026Keywords:
Catboost, Ordered Target Encoding, Weight of EvidenceAbstract
The transformation of categorical variables as part of the feature engineering process is a critical stage in classification modeling because it determines the quality of the information representation used throughout model development. This issue is particularly relevant to the development of a compliance scoring model for excise taxpayers producing or importing alcoholic beverages in Indonesia, where all predictor variables are categorical. The compliance scoring model currently implemented by the Directorate General of Customs and Excise (DGCE) is based on a manually constructed scorecard in which the categorization of predictor variables is determined subjectively. Moreover, the existing model has not been empirically validated to represent the probability of compliance violations. This study aims to evaluate the current excise compliance scoring model while developing data-driven alternatives using logistic regression and the Categorical Boosting (CatBoost) algorithm with innovative categorical encoding strategies. Two encoding methods are compared: Weight of Evidence (WoE), representing a conventional statistical approach, and Ordered Target (OT), which is adapted from the CatBoost encoding mechanism. All models are validated using repeated k-fold cross-validation and evaluated based on the Area Under the Receiver Operating Characteristic Curve (AUC), the Kolmogorov–Smirnov (KS) statistic, and several additional performance metrics. The results demonstrate that OT encoding consistently outperforms WoE across all logistic regression scenarios, increasing the KS statistic from 0.27 for the existing compliance scoring model to 0.34 with WoE and 0.42 with OT. Furthermore, the CatBoost model achieves superior predictive performance compared with both logistic regression models (AUC = 0.715; KS = 0.471), indicating stronger discriminative ability in identifying compliance violations. These findings demonstrate that integrating target-based encoding with boosting algorithms produces a more adaptive and accurate compliance scoring model, providing a robust analytical framework to support risk-based excise supervision in Indonesia.
Downloads
References
[1] Williem, C. A., & Julitawaty, W. (2024). Peran kebijakan ekstensifikasi cukai dalam melindungi kesehatan masyarakat. Jurnal Akuntansi, Keuangan, Perpajakan dan Tata Kelola Perusahaan, 1(3). https://doi.org/10.59407/jakpt.v1i3.579
[2] Direktorat Jenderal Bea dan Cukai. (2024). Instruksi Nomor INS-3/BC/2024 tentang Profil Risiko Pengusaha Barang Kena Cukai dalam Rangka Pengawasan dan Pelayanan di Bidang Cukai. Kementerian Keuangan Republik Indonesia.
[3] Margeloiu, A. (2025). Tabular Machine Learning on Small-Size and High-Dimensional Data. University of Cambridge Repository. https://doi.org/10.17863/CAM.117151
[4] Chen, K., Zhu, K., Meng, Y., Yadav, A., & Khan, A. (2019). Mixed credit scoring model of logistic regression and evidence weight in the background of big data. In J. Xu, M. Gen, & F. L. Cooke (Eds.), Advances in Intelligent Systems and Computing (Vol. 866, pp. 335–345). Springer. https://doi.org/10.1007/978-3-030-16657-1_40
[5] Siddiqi, N. (2006). Credit risk scorecards: Developing and implementing intelligent credit scoring. Hoboken, NJ: John Wiley & Sons, Inc.
[6] Duong, B. T., Sariyev, O., & Zeller, M. (2025). Can Weight of Evidence Method Improve Poverty Targeting? Social Indicators Research. https://doi.org/10.1007/s11205-025-03576-z
[7] Pomazanov, M. (2023). Continuous transformation of nonmonotonic factors of the scoring model increasing its discriminating power. Procedia Computer Science, 221, 237–244.
[8] Hancock, J. T., & Khoshgoftaar, T. M. (2020). CatBoost for Big Data: An Interdisciplinary Review. Journal of Big Data, 7(1), 1–45. https://doi.org/10.1186/s40537-020-00369-8
[9] Prokhorenkova, L., Gusev, G., Vorobev, A., Dorogush, A. V., & Gulin, A. (2018). CatBoost: Unbiased boosting with categorical features. Advances in Neural Information Processing Systems, 31. https://arxiv.org/abs/1706.09516
[10] Biarnes, A. (2021). How CatBoost encodes categorical variables? Medium. https://medium.com/data-science/how-catboost-encodes-categorical-variables-3866fb2ae640
[11] Chen, H., & Xiang, Y. (2017). The study of credit scoring model based on group lasso. Procedia Computer Science, 122, 677–684. https://doi.org/10.1016/j.procs.2017.11.423
[12] James, G., Witten, D., Hastie, T., & Tibshirani, R. (2023). An introduction to statistical learning: With applications in R and Python (3rd ed.). Springer.
[13] Persson, R. (2021). Weight of evidence transformation in credit scoring models: How does it affect the discriminatory power? [Master’s thesis, Lund University, Department of Statistics]. LUP Student Papers. https://lup.lub.lu.se/student-papers/record/9066332
[14] Kritzinger, N. & Van Vuuren, G.W. (2018). An optimised credit scorecard to enhance cut-off score determination. South African Journal of Economic and ManagementSciences 21(1).https://doi.org/10.4102/sajems.v21i1.1571
[15] Abellana, D. P. M., & Lao, D. M. M. (2023). A new univariate feature selection algorithm based on the best–worst multi-attribute decision-making method. Decision Analytics Journal, 7, 100240. https://doi.org/10.1016/j.dajour.2023.100240
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Departemen Matematika dan Sains Data FMIPA UNAND
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
All articles published in Jurnal Matematika UNAND (JMUA) are open access and licensed under the Creative Commons Attribution-ShareAlike (CC BY-SA) license. This ensures that the content is freely available to all users and can be shared and adapted, provided appropriate credit is given and any adaptations are distributed under the same license.
Copyright Holder
The copyright of all articles published in Jurnal Matematika UNAND is held by the Departemen Matematika dan Sains Data, Fakultas Matematika dan Ilmu Pengetahuan Alam (FMIPA), Universitas Andalas (UNAND). This applies to all published versions, including the HTML and PDF formats of the articles.
Author Rights
While the Departemen Matematika dan Sains Data FMIPA UNAND holds the copyright for all published content, authors retain important rights under the Creative Commons Attribution-ShareAlike 4.0 International License (CC BY-SA). This license grants authors and users the following rights:
- Reuse: Authors can reuse and distribute their work for any lawful purpose, including sharing on personal websites, institutional repositories, or in subsequent publications.
- Attribution and Adaptation: Authors and others may remix, adapt, and build upon the published work for any purpose, even commercially, as long as proper credit is given to the original authors, and any derivative works are distributed under the same CC BY-SA license.
Creative Commons License (CC BY-SA)
Under the terms of the CC BY-SA license, users are free to:
- Share: Copy and redistribute the material in any medium or format.
- Adapt: Remix, transform, and build upon the material for any purpose, even commercially.
However, the following conditions apply:
- Attribution: Users must give appropriate credit to the original author(s) and Departemen Matematika dan Sains Data FMIPA UNAND, provide a link to the license, and indicate if changes were made. Attribution must not imply endorsement by the author or the journal.
- ShareAlike: If users remix, transform, or build upon the material, they must distribute their contributions under the same license as the original.
For more information about the CC BY-SA license, please visit the Creative Commons website.
Third-Party Content
If authors include third-party material (such as figures, tables, or images) that is not covered by a Creative Commons license, they must obtain the necessary permissions for reuse and provide proper attribution. Authors are required to ensure that any third-party content complies with open-access licensing requirements or includes permissions for redistribution under similar terms.
Copyright and Licensing Information Display
The copyright and licensing terms will be clearly displayed on each article's landing page, as well as within the full-text versions (HTML and PDF) of all published articles.
No "All Rights Reserved"
As an open-access journal, JMUA does not use "All Rights Reserved" policies. Instead, the CC BY-SA license ensures that the works remain accessible and reusable for a wide audience while still protecting both the authors' and the copyright holder's rights.
Â
