Abstract

The building permit levy is a critical source of regional own-source revenue for local governments in Indonesia. Accurate prediction of levy amounts is essential for fiscal planning, transparency, and administrative efficiency. This study applies seven machine learning regression algorithms — Elastic Net, Random Forest, XGBoost, LightGBM, Gradient Boosting, Decision Tree, and k-Nearest Neighbors — to predict IMB levy values using building-related features derived from 14,971 permit application records. A comprehensive exploratory data analysis (EDA) pipeline was implemented, including missing-value audit, outlier analysis (IQR and Z-score), Pearson and Spearman correlation analysis, and data-leakage auditing. The evaluation combines a held-out test set, 5-fold cross-validation, bootstrap confidence intervals, feature ablation, raw-versus-log target comparison, randomized hyperparameter tuning, and a scenario-based analysis across levy magnitude, building function, designation, and building size segments. Random Forest achieves R² = 0.6875, MAE = Rp 5,002,478, and MAPE = 11.18% on the held-out set, while Gradient Boosting attains the best hold-out R² (0.8490); cross-validation shows the two ensembles are statistically close under the heavy-tailed target, and Elastic Net fails catastrophically (R² = −160.65), confirming the non-linear levy–feature relationship. Feature importance and ablation identify total building area (total_luas) as the overwhelmingly dominant predictor (83.76% importance), consistent with the multiplicative structure of the IMB levy formula. Scenario analysis reveals that relative errors concentrate in non-residential permits and at the extremes of the levy distribution, motivating a human-in-the-loop deployment for those segments. These findings establish tree-based ensembles as robust candidates for automated levy estimation in regional government digital services.

References

• Breiman, L. 2001. Random Forests. Machine Learning, 45(1), 5–32. https://doi.org/10.1023/A:1010933404324
• Chen, T. and Guestrin, C. 2016. XGBoost: A Scalable Tree Boosting System. In Proc. 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD ’16), 785–794. https://doi.org/10.1145/2939672.2939785
• Ke, G., Meng, Q., Finley, T., Wang, T., Chen, W., Ma, W., Ye, Q., and Liu, T.-Y. 2017. LightGBM: A Highly Efficient Gradient Boosting Decision Tree. Advances in Neural Information Processing Systems (NeurIPS), 30, 3146–3154.
• Zou, H. and Hastie, T. 2005. Regularization and Variable Selection via the Elastic Net. Journal of the Royal Statistical Society: Series B, 67(2), 301–320. https://doi.org/10.1111/j.1467-9868.2005.00503.x
• Republic of Indonesia. 2009. Undang-Undang Nomor 28 Tahun 2009 tentang Pajak Daerah dan Retribusi Daerah. Lembaran Negara RI Tahun 2009 Nomor 130. Jakarta: Sekretariat Negara.
• Bentejac, C., Csorgo, A., and Martinez-Munoz, G. 2021. A Comparative Analysis of Gradient Boosting Algorithms. Artificial Intelligence Review, 54(3), 1937–1967. https://doi.org/10.1007/s10462-020-09896-5
• Grinsztajn, L., Oyallon, E., and Varoquaux, G. 2022. Why Do Tree-Based Models Still Outperform Deep Learning on Tabular Data? Advances in Neural Information Processing Systems (NeurIPS 2022), 35. https://doi.org/10.48550/arXiv.2207.08815
• Mayer, M., Bourassa, S. C., Hoesli, M., and Scognamiglio, D. 2022. Machine Learning Applications to Land and Structure Valuation. Journal of Risk and Financial Management, 15(5), 193. https://doi.org/10.3390/jrfm15050193
• Zilli, C. A. and Bastos, L. C. 2024. Machine Learning Models in Mass Appraisal for Property Tax Purposes: A Systematic Mapping Study. Aestimum, 84, 31–52. https://doi.org/10.36253/aestim-15792
• Sharma, H., Harsora, H., and Ogunleye, B. 2024. An Optimal House Price Prediction Algorithm: XGBoost. Analytics, 3(1), 30–45. https://doi.org/10.3390/analytics3010003
• Xu, C. and Kong, Y. 2024. Random Forest Model in Tax Risk Identification of Real Estate Enterprise Income Tax. PLOS ONE, 19(3), e0300928. https://doi.org/10.1371/journal.pone.0300928
• Lundberg, S. M. and Lee, S.-I. 2017. A Unified Approach to Interpreting Model Predictions. Advances in Neural Information Processing Systems (NeurIPS), 30, 4765–4774.
• Apriliyanti, I. D., Kusumasari, B., Pramusinto, A., and Setianto, W. A. 2021. Digital Divide in ASEAN Member States: Analyzing the Critical Factors for Successful E-Government Programs. Online Information Review, 45(2), 440–460. https://doi.org/10.1108/OIR-03-2020-0093
• Dwivedi, Y. K., Rana, N. P., Janssen, M., Lal, B., Williams, M. D., and Clement, M. 2023. An Empirical Validation of a Unified Model of Electronic Government Adoption (UMEGA). Government Information Quarterly, 40(2), 101819. https://doi.org/10.1016/j.giq.2022.101819
• Zilli, C. A., Bastos, L. C., and Da Silva, L. R. 2024. Mass Appraisal of Apartments Using Random Forest and Gradient Boosting Algorithms: Case Study of Florianopolis, Brazil. Revista do Departamento de Geografia, 44, e212297. https://doi.org/10.11606/eISSN.2236-2878.rdg.2024.212297
• Antipov, E. A. and Pokryshevskaya, E. B. 2023. Boosting the Accuracy of Commercial Real Estate Appraisals: An Interpretable Machine Learning Approach. The Journal of Real Estate Finance and Economics. https://doi.org/10.1007/s11146-023-09944-1
• Yang, L. 2025. Predictive Modeling of Tax Compliance Risks: A Comparative Study of Machine Learning Approaches. PLOS ONE, 20(9), e0331715. https://doi.org/10.1371/journal.pone.0331715
• Moreno-Lopez, M., Zofio, J. L., and Gonzalez-Arias, J. 2025. Comparative Analysis of Advanced Models for Predicting Housing Prices: A Review. Buildings, 9(2), 32. https://doi.org/10.3390/urbansci9020032
• Indah, Y. M., Sumarni, S., and Widiyastuti, T. 2024. Comparison of Random Forest, XGBoost and LightGBM Methods on the Human Development Index in Indonesia. Jurnal Sains dan Informatika, 10(2), 155–164. https://doi.org/10.34128/jsi.v10i2.651
• Yilmaz, A. K. and Kalayci, A. 2024. Gradient Boosting Decision Trees on Medical Diagnosis over Tabular Data. In Proceedings of the 2024 IEEE International Conference on Big Data,4521–4527. https://doi.org/10.1109/ICAD65464.2025.11114069
• Kepplinger, D. 2023. Robust Variable Selection and Estimation via Adaptive Elastic Net S-Estimators for Linear Regression. Computational Statistics & Data Analysis, 183, 107730. https://doi.org/10.1016/j.csda.2023.107730
• Tsagkis, P., Bakogiannis, E., and Nikitas, A. 2023. Analysing Urban Growth Using Machine Learning and Open Data: An Artificial Neural Network Modelled Case Study of Five Greek Cities. Sustainable Cities and Society, 89, 104337. https://doi.org/10.1016/j.scs.2022.104337
• Shwartz-Ziv, R. and Armon, A. 2022. Tabular Data: Deep Learning Is Not All You Need. Information Fusion, 81, 84–90. https://doi.org/10.1016/j.inffus.2021.11.011
• Firmansyah, R. and Wicaksono, A. 2021. E-Government Implementation and Corruption Reduction: Evidence from Indonesian Local Governments. Asian Journal of Political Science, 29(4), 445–463. https://doi.org/10.1080/02185377.2021.1978143
• Shaputra, E., Ginting, B. S., and Nurhayati, N. 2021. Prediksi Pendapatan Asli Daerah (PAD) Kabupaten Langkat Menggunakan Metode Backpropagation Neural Network. JTIK (Jurnal Teknik Informatika Kaputama), 5(1), 69–75. https://doi.org/10.59697/jtik.v5i1.587
• Nagappan, M. and Daud, S. M. 2021. Machine Learning Predictors for Sustainable Urban Planning. International Journal of Advanced Computer Science and Applications (IJACSA), 12(7), 457–464. https://doi.org/10.14569/IJACSA.2021.0120754
• Republic of Indonesia. 2021. Peraturan Pemerintah Nomor 16 Tahun 2021 tentang Peraturan Pelaksanaan Undang-Undang Nomor 28 Tahun 2002 tentang Bangunan Gedung. Lembaran Negara RI Tahun 2021 Nomor 26. Jakarta: Sekretariat Negara.
• Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., et al. 2011. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research, 12, 2825–2830. (Scopus-indexed; widely cited in ML pipeline implementations)
• Robi, R. K. and George, J. K. 2025. Application of Machine Learning Algorithms to Predict Urban Expansion. Journal of Urban Planning and Development, 151(2). https://doi.org/10.1061/JUPDDM.UPENG-5466
• Hachcha, F. and Kessentini, H. 2022. Elastic Net Penalized Quantile Regression Model. Journal of Computational and Applied Mathematics, 409, 113800. https://doi.org/10.1016/j.cam.2022.113800