Sign in Subscribe
Energy management

Advanced Machine Learning Approaches for Cooling Load Prediction in Residential Buildings

Strategic Insights

— 1 min read

Researchers at Concordia University in Montreal, Quebec, Canada, have made significant progress in developing a novel approach for predicting cooling loads in residential buildings using advanced machine learning techniques. The study, published in the Journal of Engineering, exploited the potential of ensemble methods combined with optimization-driven parameter tuning to improve the accuracy and robustness of cooling load prediction. The researchers proposed a hybrid framework, Voting Regression (VR) with Weevil Damage Optimization (WDO) and Dandelion Sowing Optimization (DSO), which achieved outstanding performance in predicting cooling loads.

Key Takeaways:

  • The proposed hybrid framework, VR-WDO, recorded an R² of 0.988 and RMSE of 1.034 on the training dataset and an R² of 0.983 with RMSE of 1.209 on the test dataset.
  • The CatBoost Regression (CATR) model yielded the weakest performance across the evaluation metrics, highlighting the superiority of the proposed hybrid approach.
  • The study demonstrated that combining ensemble methods with optimization-driven parameter tuning addresses limitations in conventional models and offers practical insights for energy-efficient building design.
  • The results provide a reliable decision-support tool for engineers, architects, and urban planners seeking to enhance energy planning and optimize HVAC system performance in residential environments.
  • The developed framework uses critical factors such as building envelope characteristics, solar exposure, and internal heat sources to forecast cooling demand.
  • The study introduced a novel ensemble framework, Voting Regression (VR), and enhanced it through integration with metaheuristic optimization algorithms WDO and DSO.

Statistics:

  • R² of 0.988 on the training dataset for the VR-WDO model
  • RMSE of 1.034 on the training dataset for the VR-WDO model
  • R² of 0.983 on the test dataset for the VR-WDO model
  • RMSE of 1.209 on the test dataset for the VR-WDO model

Sources:

  • NewsRx. Data from Concordia University Advance Knowledge in Machine Learning (Cooling Load Prediction in Residential Buildings: Machine Learning Approaches and Comparative Analysis). Journal of Engineering. October 20, 2025; p 216.
  • Cooling Load Prediction in Residential Buildings: Machine Learning Approaches and Comparative Analysis. Advances in Engineering and Intelligence Systems, 2025,004(03):86-100.

Read more