THERMAL SCIENCE

International Scientific Journal

Xingzhe Zhu

,

Di Zhou

,

Wenyuan Wang

,

Zebiao Shao

ENERGY PERFORMANCE PREDICTION AND INTERPRETABILITY ANALYSIS OF RESIDENTIAL BUILDING COMPLEXES USING AUTOMATED MACHINE LEARNING

ABSTRACT
Data-driven models for predicting the energy performance of residential complexes often rely on expert knowledge, making the modeling process time-consuming. This study presents a predictive framework that integrates automated machine learning (AutoML) with interpretability analysis. A typical parallel-arranged residential complex in Tianjin was used as a case study to forecast both energy use intensity (EUI) and photovoltaic generation potential (PVG). A dataset encompassing building morphology, layout, and envelope parameters was constructed, and six representative AutoML platforms (TPOT, FLAML, MLjar, AutoGluon, LightAutoML, and H2O AutoML) were employed to develop predictive models. The results indicate that all frameworks achieved high predictive accuracy, although performance varied: LightAutoML performed best for EUI prediction, MLjar for PVG prediction, and FLAML exhibited consistent performance across both tasks. Interpretability analyses-utilizing variable importance metrics, SHAP values, partial dependence (PD), and individual conditional expectation (ICE) curves-revealed that EUI is primarily influenced by floor height, building length, solar heat gain coefficient, and north-south window-to-wall ratio, whereas PVG is mainly affected by the number of floors, building spacing, building length, and building height. This framework enables rapid prediction using readily available design parameters and reveals both qualitative and quantitative relationships between key variables and energy performance, providing support for early-stage low-carbon planning and design optimization in residential complexes.
KEYWORDS
automated machine learning, model interpretability, building energy intensity, photovoltaic potential, influencing factors
PAPER SUBMITTED: 2026-05-10
PAPER REVISED: 2026-07-07
PAPER ACCEPTED: 2026-07-11
PUBLISHED ONLINE: 2026-09-03
DOI REFERENCE: https://doi.org/10.2298/TSCI260510167Z
REFERENCES (44)
[1] Keramidas, K., et al., Global Energy and Climate Outlook 2020: A New Normal Beyond Covid-19, Publications Office of the European Union, Luxembourg, 2021
[2] Hu, S., Jiang, Y., China Building Energy Use and Carbon Emission Yearbook 2021: A Roadmap to Carbon Neutrality by 2060, Springer, 2022
[3] Yuan, P., et al., Analysis of indoor thermal environment and human thermal comfort in rural residences of Northern Henan, China, Thermal Science, (2026), pp. 18-18
[4] Ali, U., et al., Urban building energy performance prediction and retrofit analysis using data-driven machine learning approach, Energy and Buildings, 303 (2024), 113768
[5] Tehrani, A. A., et al., Data-driven approach to estimate urban heat island impacts on building energy consumption, Energy, 316 (2025), 134508
[6] Yıldız, M. E., Beyhan, F., Prediction of cooling load via machine learning on building envelope design parameters, Journal of Building Engineering, 100 (2025), 111724
[7] Wang, W., et al., Integrating local climate considerations into dormitory energy optimization: An explainable machine learning and multi-objective design approach, Urban Climate, 64 (2025), 102629
[8] Wolpert, D. H., Macready, W. G., No free lunch theorems for optimization, IEEE Transactions on Evolutionary Computation, 1(1) (2002), pp. 67-82
[9] Li, D., et al., Machine learning applications in building energy systems: review and prospects, Buildings, 15(4) (2025), 648
[10] Shafiuzzaman, M., et al., Enhanced very short-term load forecasting with multi-lag feature engineering and prophet-XGBoost-CatBoost architecture, Energy, 137981 (2025)
[11] Koukaras, P., Tjortjis, C., Data Preprocessing and Feature Engineering for Data Mining: Techniques, Tools, and Best Practices, AI, 6(10) (2025)
[12] Cho, J. H., Moon, J. W., Integrated artificial neural network prediction model of indoor environmental quality in a school building, Journal of Cleaner Production, 344 (2022), 131083
[13] Cai, W., et al., Predicting the energy consumption in buildings using the optimized support vector regression model, Energy, 273 (2023), 127188
[14] Mohammadagha, M., Hyperparameter optimization strategies for tree-based machine learning models prediction: a comparative study of AdaBoost, decision trees, and random forest, Decision Trees, and Random Forest, April 11, 2025
[15] Kaloop, M. R., et al., Predicting energy consumption of residential buildings using metaheuristic-optimized artificial neural network technique in early design stage, Building and Environment, 274 (2025), 112749
[16] Xu, W., et al., Predicting daily heating energy consumption in residential buildings through integration of random forest model and meta-heuristic algorithms, Energy, 301 (2024), 131726
[17] Mu, W., et al., Data Preprocessing Techniques for Machine Learning Towards Improving Building Energy Performance: A Systematic Review, Energies, 19(6) (2026), 1561
[18] Singh, D., Singh, B., Feature wise normalization: An effective way of normalizing data, Pattern Recognition, 122 (2022), 108307
[19] Mishra, D., et al., A review on ensemble learning methods: machine learning approach, International Journal of Research Publication and Reviews, 6 (2025), 2, pp. 3795-3803
[20] Wang, Z., et al., A comprehensive review and future research directions of ensemble learning models for predicting building energy consumption, Energy and Buildings, 335 (2025), 115589
[21] Vadar, P. S., et al., Comparative analysis of automated machine learning libraries: pycaret, h2o, tpot, auto-sklearn, and flaml, Int. J. Sci. Res. Eng. Manag, 8(11) (2024), 1-8
[22] Anitha, M., et al., AutoML Ecosystem and Open-Source Platforms: Challenges and Limitations, Automated Machine Learning and Industrial Applications, 191-205 (2025)
[23] Wang, C., et al., Automated machine learning & tuning with flaml, in: Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, August 2022, pp. 4828-4829
[24] Daniel, C., Neufville, E. K., An AutoGluon-enabled robust machine learning model for concrete tensile and compressive strength forecast, International Journal of Construction Management, 25(13) (2025), 1636-1647
[25] Manga, I., AutoML for All: Democratizing Machine Learning Model Building with Minimal Code Interfaces, in: Proceedings of the 3rd International Conference on Sustainable Computing and Data Communication Systems (ICSCDS), August 2025, pp. 347-352, IEEE
[26] Zhang, C., et al., Automated machine learning-based building energy load prediction method, Journal of Building Engineering, 80 (2023), 108071
[27] Baratchi, M., et al., Automated machine learning: past, present and future, Artificial Intelligence Review, 57(5) (2024), 122
[28] Papadopoulos, D. N., et al., Harnessing AutoML for Electricity Price Forecasting in Germany: A Comparative Analysis of Models, in: Proceedings of the 21st International Conference on the European Energy Market (EEM), May 2025, pp. 1-6, IEEE
[29] Ren, Y., Yu, X., Forecasting renewable energy generation based on AutoML and transfer learning, in: IET Conference Proceedings CP892, Vol. 2024, No. 14, May 2024, pp. 1-7, Stevenage, UK: The Institution of Engineering and Technology
[30] Zhang, C., et al., Intrinsically interpretable machine learning-based building energy load prediction method with high accuracy and strong interpretability, Energy and Built Environment, 2024
[31] Ang, Y. Q., et al., On interpretable and explainable machine learning for urban building energy modeling, in: Proceedings of the 15th ACM International Conference on Future and Sustainable Energy Systems, June 2024, pp. 683-686
[32] Chen, Z., et al., Interpretable machine learning for building energy management: A state-of-the-art review, Advances in Applied Energy, 9 (2023), 100123
[33] Abdeyazdan, H., et al., Explainable machine learning for passive design: Early-stage building energy reduction in hot-arid climates, Sustainable Energy Technologies and Assessments, 83 (2025), 104589
[34] Li, X., et al., Building energy prediction in a changing climate: an interpretable machine learning approach, Building and Environment, 113420 (2025)
[35] Singh, V. K., Joshi, K., Automated machine learning (AutoML): an overview of opportunities for application and research, Journal of Information Technology Case and Application Research, 24(2) (2022), 75-85
[36] Sun, N., et al., Construction of Automated Machine Learning (AutoML) Framework Based on Large Language Models, in: Proceedings of the 6th International Conference on Computer Engineering and Application (ICCEA), April 2025, pp. 1-4, IEEE
[37] Shujaat, S., Automated machine learning in dentistry: A narrative review of applications, challenges, and future directions, Diagnostics, 15(3) (2025), 273
[38] Ministry of Housing and Urban-Rural Development of the People's Republic of China, Code of urban Residential Areas Planning and Design (GB50180-2018), Beijing, 2018
[39] Wang, P., et al., Sustainability of compact cities: A review of inter-building effect on building energy and solar energy use, Sustainable Cities and Society, 72 (2021), 103035
[40] Larson, G. W., et al., Rendering with Radiance, Morgan Kaufmann Publishers, San Francisco, CA, USA, 1998
[41] Lundberg, S. M., Lee, S.-I., A unified approach to interpreting model predictions, Advances in Neural Information Processing Systems, 30 (2017)
[42] Makowski, D., et al., modelbased: An R package to make the most out of your statistical models through marginal means, marginal effects, and model predictions, Journal of Open Source Software, 10(109) (2025), 7969
[43] Tian, J., Ooka, R., Evaluation of solar energy potential for residential buildings in urban environments based on a parametric approach, Sustainable Cities and Society, 106 (2024), 105350
[44] Xu, S., et al., Influence of urban morphological factors on building energy consumption combined with photovoltaic potential: A case study of residential blocks in central China, in: Building Simulation, 16(9) (2023), pp. 1777-1792, Beijing: Tsinghua University Press
PDF VERSION [DOWNLOAD]
Energy performance prediction and interpretability analysis of residential building complexes using automated machine learning

© 2026 Society of Thermal Engineers of Serbia. Published by the Vinča Institute of Nuclear Sciences, National Institute of the Republic of Serbia, Belgrade, Serbia. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution 4.0 International licence