Improving Energy Efficiency in Building Operations through Enhanced Cooling Load Prediction using Decision Tree Model and Dual Meta-Heuristic Optimizations

Predicting the Cooling Load (CL) of a building’s energy consumption is paramount in optimizing energy usage and mitigating environmental impacts. This study implemented a Decision Tree (DT) model to forecast the CL, aiming to optimize energy efficiency in building operations. To refine the predictive accuracy and generalization capability of the DT model, two advanced metaheuristic optimization algorithms were employed: the Giant Trevally Optimizer (GTO) and the Equilibrium Slime Mould Algorithm (ESMA). The algorithms were pivotal in hyperparameter tuning and model optimization, leading to enhanced performance metrics and more robust predictive outcomes. In this article, the models are evaluated using several metrics, including the Scatter Index (SI), Correlation Coefficient (R²), Mean Absolute Error (MAE), BIAS, and Root Mean Square Error (RMSE), where strong algorithm performance is determined by a high R² value and reduced model error is preferred with lower RMSE and MAE values. The findings yielded promising outcomes. The DTGT model, incorporating the GTO optimizer, exhibited remarkable performance, boasting an exceptional R2 value of 0.996, indicating a near-perfect alignment with the dataset. Moreover, the Root Mean Square Error (RMSE) value of 0.604 underscored the model’s exceptional precision in making predictions, with an impressively low error margin. These results underscore the resilience of the DTGT model as a useful instrument for exceptionally precise CL forecasts. Effective prediction of the CL leads to enhancing energy efficiency, reducing operational costs, and supporting smart building technologies by providing accurate.