Machine-learning-based assessment of heating and cooling degree days for predicting building energy demand
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
The authors constructed a model to predict future energy consumption in the Bukhara region using two meteorological indicators, heating degree days and cooling degree days. Using daily temperature data from Climate-Data, Timeanddate, and Meteoblue, the authors created a dataset and calculated heating and cooling degree days for each day, revealing seasonal variation in energy usage. The authors used three methods to build a regression model: random forest regression, XGBoost regression, and linear regression. The evaluation results clearly demonstrate that all three regression models accurately predicted future energy consumption. The regression models had an RMSE of 8.13, an MAE of 6.43, an R² of 0.944, and an MAPE of 2.65%. The authors concluded from their study that heating degree days were the primary indicator for energy utilisation in the cold months. At the same time, cooling degree days are a vital indicator for summer and significantly influence cooling load development. The methodology adopted by the authors will be another tool to strengthen their predicted models of energy use, assess the building's insulation capabilities, and optimise and manage building energy use.
Fields of Study & Subjects
Building Energy and Comfort Optimization
Energy Load and Power Forecasting
Building energy efficiency and sustainability
Publication details
- Journal / source
- International Conference on Optics, Materials Science, and Computing in Condensed-Matter Physics (OMCCP)
- Pages
- 62–62
- Publication date
- 30 March 2026
- Language
- EN