Journals / International Journal of Multidisciplinary Studies and Innovative Technologies / 2026 / Cilt: 10 Sayı: 1

Cross-Regional Methodological Validation of Ensemble and Kernel-Based Machine Learning Models for Automated Building Energy Performance Assessment

Bölgeler Arası Topluluk ve Çekirdek Tabanlı Makine Öğrenmesi Modellerinin Otomatik Bina Enerji Performansı Değerlendirmesi İçin Metodolojik Doğrulaması

Pages
102–116
Publication Date
2026-08-31
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

Unlike earlier single-dataset studies, this work removes optical character recognition (OCR) document ingestion from the scope and instead concentrates on a rigorous, side-by-side comparison of four learning paradigms across independent regional datasets, offering evidence for the predictive ceiling of ensemble learning on tabular energy data and for the conditions under which a smaller, locally collected dataset can achieve results comparable to those from a larger but less homogeneous metropolitan dataset.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:This paper presents an extended machine learning framework for predicting building Energy Performance Certificate (EPC) scores and evaluates its consistency across regulatory and geographic contexts.
📊
Key Findings & Data:Building on two prior single-context studies, the present work integrates four regression paradigms — SVR, Random Forest, XGBoost, and, AdaBoost — trained on the Seattle Building Energy Benchmarking dataset (34,709 records, 46 parameters) and evaluated against real municipal EPC and building-registry data collected from three Turkish administrative districts: Torbalı, Beşiktaş, and Ümraniye.
💡
Conclusion & Impact:Unlike earlier single-dataset studies, this work removes optical character recognition (OCR) document ingestion from the scope and instead concentrates on a rigorous, side-by-side comparison of four learning paradigms across independent regional datasets, offering evidence for the predictive ceiling of ensemble learning on tabular energy data and for the conditions under which a smaller, locally collected dataset can achieve results comparable to those from a larger but less homogeneous metropolitan dataset.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Enerji Performans Sertifikası (EPS) · makine öğrenmesi · Destek Vektör Regresyonu

Abstract

This paper presents an extended machine learning framework for predicting building Energy Performance Certificate (EPC) scores and evaluates its consistency across regulatory and geographic contexts. Building on two prior single-context studies, the present work integrates four regression paradigms — SVR, Random Forest, XGBoost, and, AdaBoost — trained on the Seattle Building Energy Benchmarking dataset (34,709 records, 46 parameters) and evaluated against real municipal EPC and building-registry data collected from three Turkish administrative districts: Torbalı, Beşiktaş, and Ümraniye. A structured preprocessing pipeline addressing missing data, outlier treatment, correlation-driven feature reduction, and categorical encoding was applied consistently across all data sources. Among the four models, the tree-based ensembles outperformed both the kernel-based SVR and the boosting-based AdaBoost on both datasets, with XGBoost achieving the strongest fit on the Seattle benchmark (R² = 0.801) and again the strongest fit on the Torbalı municipal sample (R² = 0.824). Exploratory analysis of the Turkish municipal data further identifies insulation status, construction era, and heating-fuel type as the dominant levers for emissions reduction, with uninsulated buildings consuming approximately 71% more site energy than fully insulated stock. Unlike earlier single-dataset studies, this work removes optical character recognition (OCR) document ingestion from the scope and instead concentrates on a rigorous, side-by-side comparison of four learning paradigms across independent regional datasets, offering evidence for the predictive ceiling of ensemble learning on tabular energy data and for the conditions under which a smaller, locally collected dataset can achieve results comparable to those from a larger but less homogeneous metropolitan dataset.

Özet

Bu çalışma, binaların Enerji Performans Sertifikası (EPS) puanlarının tahmini için geliştirilmiş bir makine öğrenmesi çerçevesi sunmakta ve bu çerçevenin farklı coğrafi ve düzenleyici bağlamlardaki tutarlılığını değerlendirmektedir. Önceki iki tek bağlamlı çalışmayı temel alan bu araştırmada, Seattle Bina Enerji Kıyaslama veri kümesi (34.709 kayıt, 46 özellik) üzerinde eğitilen ve Türkiye'deki üç farklı idari ilçeden (Torbalı, Beşiktaş ve Ümraniye) toplanan gerçek belediye EPS ve bina envanteri verileri üzerinde test edilen dört regresyon yaklaşımı (Destek Vektör Regresyonu (SVR), Random Forest, XGBoost ve AdaBoost) karşılaştırılmıştır. Tüm veri kaynaklarına eksik verilerin giderilmesi, aykırı değerlerin işlenmesi, korelasyona dayalı özellik azaltımı ve kategorik değişken kodlamasını içeren yapılandırılmış bir ön işleme süreci tutarlı biçimde uygulanmıştır. Dört model arasında ağaç tabanlı topluluk yöntemleri, hem çekirdek tabanlı SVR'yi hem de artırmalı öğrenme yöntemi olan AdaBoost'u her iki veri kümesinde de geride bırakmıştır. XGBoost modeli, hem Seattle kıyaslama veri kümesinde (R² = 0.801) hem de Torbalı belediye örnekleminde (R² = 0.824) en yüksek tahmin başarısını elde etmiştir. Türk belediye verileri üzerinde gerçekleştirilen keşifsel analizler, yalıtım durumu, bina yapım dönemi ve ısınma yakıt türünün emisyonların azaltılmasında belirleyici faktörler olduğunu göstermiş; yalıtımsız binaların tam yalıtımlı binalara kıyasla yaklaşık %71 daha fazla yerinde enerji tükettiği belirlenmiştir. Önceki tek veri kümesine dayalı çalışmalardan farklı olarak bu araştırma, optik karakter tanıma (OCR) tabanlı belge işleme sürecini kapsam dışı bırakmış ve bunun yerine dört farklı makine öğrenmesi yaklaşımının bağımsız bölgesel veri kümeleri üzerindeki performanslarını sistematik olarak karşılaştırmaya odaklanmıştır. Elde edilen bulgular, topluluk öğrenmesi yöntemlerinin tablo tabanlı enerji verilerindeki öngörü kapasitesini ve daha küçük ancak yerel olarak toplanmış veri kümelerinin, daha büyük fakat daha heterojen metropol veri kümeleriyle karşılaştırılabilir doğruluk düzeyine hangi koşullar altında ulaşabileceğini ortaya koymaktadır.

🔖

Cite This Article

Tanrıbuyruğu, T., Olca, E. (2026). Cross-Regional Methodological Validation of Ensemble and Kernel-Based Machine Learning Models for Automated Building Energy Performance Assessment. International Journal of Multidisciplinary Studies and Innovative Technologies, Cilt: 10 Sayı: 1, 102–116. https://doi.org/10.36287/ijmsit.10.1.12

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (28)

  1. European Commission, "Energy Performance of Buildings Directive," Directorate-General for Energy, European Commission. [Online]. Available: https://energy.ec.europa.eu/topics/energy-efficiency/energy-performance-buildings/energy-performance-buildings-directive_en.
  2. United Nations, "The 17 Goals | Sustainable Development," United Nations Department of Economic and Social Affairs, [Online]. Available: https://sdgs.un.org/goals.
  3. D. G. Yılmaz and F. Cesur, "A Study for the Improvement of the Energy Performance Certificate (EPC) System in Turkey," Sustainability, vol. 15, no. 19, Art. no. 14074, Sep. 2023, doi: 10.3390/su151914074.
  4. Bankacılık Düzenleme ve Denetleme Kurumu (BDDK), "Bankacılık Düzenleme ve Denetleme Kurulu Kararı, Karar Sayısı: 11165, 13 Mart 2025" [Banking Regulation and Supervision Board Decision No. 11165, Mar. 13, 2025], 2025. [Online]. Available: https://www.bddk.gov.tr/Mevzuat/DokumanGetir/1285.
  5. U. Kınay and A. Laukkarinen, "Renovation wave of the residential building stock targets for the carbon-neutral: Evaluation by Finland and Türkiye case studies for energy demand," Buildings and Cities, 2023.
  6. European Parliament and Council of the European Union, Directive 2002/91/EC of the European Parliament and of the Council of 16 December 2002 on the Energy Performance of Buildings, Official Journal of the European Communities, L 1, pp. 65–71, Jan. 4, 2003.
  7. U.S. Environmental Protection Agency, ENERGY STAR Score: Technical Reference, ENERGY STAR Portfolio Manager, Apr. 2021. [Online]. Available: https://portfoliomanager.energystar.gov/pdf/reference/ENERGY%20STAR%20Score.pdf.
  8. J. Scott, "Grid Edge: Artificial Intelligence for Energy Systems," presented at IEA Workshop on Modernising Energy Efficiency through Digitalisation, Paris, France, 2019.
  9. P. Rocha, A. Siddiqui, and M. Stadler, "Improving energy efficiency via smart building energy management systems: a comparison with policy measures," Energy and Buildings, vol. 88, pp. 203-213, 2015.
  10. S. Kaur et al., "Leveraging artificial intelligence for enhanced sustainable energy management," 2024.
  11. D. Joseph and P. NS, "Leveraging generative AI for process innovation and EPC project delivery," Mintmesh Inc., 2024.
  12. S. E. Bibri, J. Krogstie, R. Kaboli, and A. Bibri, "Smarter eco-cities and their leading-edge artificial intelligence of things solutions for environmental sustainability," Energy Informatics, vol. 6, no. 1, 2023.
  13. A. J. Bokolo, " Decentralized AIoT based intelligence for sustainable energy prosumption in local energy communities: A citizen-centric prosumer approach," 2024.
  14. M. Karlı and F. Terzi, "Modelling the energy footprint of urban form in Istanbul: a big data and machine learning approach," Energy and Buildings, vol. 333, p. 115495, 2025.
  15. S. Sisman, A. Kara, and A. C. Aydınoğlu, "Leveraging spatial data infrastructure for machine learning based building energy performance prediction," PLOS One, vol. 20, no. 10, e0335531, 2025.
  16. I. Shteriyanov et al., "Smart Material Estimation for the Engineering, Procurement, and Construction (EPC) Sector," 2023.
  17. J. C. Lam, K. K. W. Wan, C. L. Tsang, and L. Yang, "Building energy efficiency in different climates," Energy Conversion and Management, vol. 49, no. 8, pp. 2354-2366, 2008.
  18. M. A. Abdillahi and T. Etem, "Stacked Ensemble Machine Learning Model of Energy Consumption Prediction," Alfa Mühendislik ve Uygulamalı Bilimler Dergisi, vol. 3, no. 2, pp. 1-14, 2025.
  19. A. Sağıroğlu and A. Taşkın, "Binalarda Enerji Verimliliğine Makine Öğrenmesi Temelli Hibrit Bir Kümeleme ve Tahminleme Yaklaşımı" [A Machine Learning-Based Hybrid Clustering and Prediction Approach to Energy Efficiency in Buildings], Karadeniz Fen Bilimleri Dergisi (The Black Sea Journal of Sciences), vol. 15, no. 4, pp. 1399–1425, Dec. 2025, doi: 10.31466/kfbd.1533972.
  20. N. Hewitt, M. Huang, B. Kamkari, and L. Darvishvand, “A systematic review of explainable artificial intelligence in urban building energy modeling: Methods, applications, and future directions,” Sustainable Cities and Society, vol. 128, Art. no. 106492, 2025, doi: 10.1016/j.scs.2025.106492.
  21. V. Arabzadeh and R. Frank, “A four-dimensional analysis of explainable AI in energy forecasting: A domain-specific systematic review,” Machine Learning and Knowledge Extraction, vol. 7, no. 4, Art. no. 153, 2025, doi: 10.3390/make7040153.
  22. N. Guo, W. Cong, Y. Su, S. Zhang, and Q. Man, “Bayesian-optimized CNN-LSTM for residential building energy consumption prediction with SHAP explainability,” Case Studies in Thermal Engineering, vol. 81, Art. no. 107870, May 2026, doi: 10.1016/j.csite.2026.107870.
  23. City of Seattle, "Seattle Building Energy Benchmarking Dataset, 2015- 2023," Open Data Portal, 2023. [Online]. Available: https://data.seattle.gov
  24. Beşiktaş Municipality, Building Energy Performance and Registry Dataset, unpublished municipal dataset provided to the authors upon request, Beşiktaş Municipality, İstanbul, Türkiye, 2025.
  25. Ümraniye Municipality, Building Energy Performance and Registry Dataset, unpublished municipal dataset provided to the authors upon request, Ümraniye Municipality, İstanbul, Türkiye, 2025.
  26. Torbalı Municipality, Building Energy Performance and Registry Dataset (2023), unpublished municipal dataset provided to the authors under a data-sharing agreement, Torbalı Municipality, İzmir, Türkiye, 2025.
  27. City of Philadelphia, "Philadelphia Building Energy Benchmarking Dataset, 2013-2023," Open Data Portal, 2023.
  28. United Kingdom Government, "Energy Performance of Buildings Data: England and Wales," gov.uk, 2023.