EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A Mathematical Decision Science Framework for Energy-Efficiency Prediction and Multi-Criteria Building Performance Assessment

Rakesh Chandra Mishra, Mr. Ratan Jana, Ankita Tiwari, Dr C Senthamaraikannan, Dr. Parvesh Kumar, Dr. Subrata Kumar Sahu

Computer and Decision Making: An International Journal · 02.11.2026 · 3 (6S) · 3429-3441

DOI: 10.1922/comdem.v3i6s.19074 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computer and Decision Making: An International Journal, online erschienen am 02.11.2026. Verfasst von Rakesh Chandra Mishra und 5 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

While accurate building energy forecasting is essential for efficient building operation, there is no certainty for the building performance comparison when the forecasting accuracy is high. A machine-learning-driven Uncertainty-Aware Predictive Multi-Criteria Decision Science (UP-MCDS) framework is proposed, which incorporates machine learning-based forecasting, SHAP explanations, conformal uncertainty estimation, and TOPSIS assessment. The framework is designed around the Building Data Genome Project 2, and assumes that 320 buildings will be sampled, and that about 5.25 million hourly observations will be made. There are six forecasting methods compared with chronological partitions and feature ablation takes an account of historical demand, weather, and temporal information. Based on the illustrative results, LightGBM can reach the lowest mean absolute error (18.92 kWh), root mean squared error (29.41 kWh) and weighted absolute percentage error (12.12%). The mean absolute error of the historical demand features is increased by 57.19% with the removal of the features. The nominal 90% conformal interval achieves a coverage of 90.4%. A multi-criteria score is used for ranking the building performance based on three factors: energy intensity, peak-demand ratio and demand variability, which are all adjusted for weather. Sensitivity analysis also explores the consistency of the rankings when using different objective weights. The framework provides an interpretable link between energy prediction and decision support, while the numerical results assumed in the framework will need to be validated and verified by external measurements to draw empirical conclusions.

Zitieren

APA

Mishra, R. C., Jana, M. R., Tiwari, A., Senthamaraikannan, D. C., Kumar, D. P., & Sahu, D. S. K. (2026). A Mathematical Decision Science Framework for Energy-Efficiency Prediction and Multi-Criteria Building Performance Assessment. Computer and Decision Making: An International Journal, 3(6S), 3429-3441. https://doi.org/10.1922/comdem.v3i6s.19074

Vancouver

Mishra RC, Jana MR, Tiwari A, Senthamaraikannan DC, Kumar DP, Sahu DSK. A Mathematical Decision Science Framework for Energy-Efficiency Prediction and Multi-Criteria Building Performance Assessment. Computer and Decision Making: An International Journal. 2026;3(6S):3429-3441. doi:10.1922/comdem.v3i6s.19074

BibTeX
@article{Mishra2026,
  author = {Mishra, Rakesh Chandra and Jana, Mr. Ratan and Tiwari, Ankita and Senthamaraikannan, Dr C and Kumar, Dr. Parvesh and Sahu, Dr. Subrata Kumar},
  title = {A Mathematical Decision Science Framework for Energy-Efficiency Prediction and Multi-Criteria Building Performance Assessment},
  journal = {Computer and Decision Making: An International Journal},
  year = {2026},
  volume = {3},
  number = {6S},
  pages = {3429--3441},
  doi = {10.1922/comdem.v3i6s.19074}
}
Kostenlos starten →