Active learning-based machine learning approach for enhancing environmental sustainability in green building energy consumption

Abstract Green building (GB) techniques are essential for reducing energy waste in the construction sector, which accounts for almost 40% of global energy consumption. Despite their importance, challenges such as occupant behavior and energy management gaps often result in GBs consuming up to 2.5 times more energy than intended. To address this, Building Automation Systems (BAS) play a crucial role in enhancing energy efficiency. This research develops a predictive model for GB design using machine learning to minimize energy consumption and improve indoor sustainability. The dataset is utilized to predict cooling and heating individually, with data visualization by graphically illustrating dataset features and preprocessing through Z-Score normalization and dataset splitting. The proposed model, based on active learning and utilizing ML regressors such as Random Forest (RF), Decision Tree (DT), Gradient Boosting (GB), Extreme Gradient Boosting (XGBoost), CatBoost (CB), Light Gradient Boosting Machine (LGBM), K-Nearest Neighbor (KNN), and Logistic Regressor (LR), shows significant performance improvements. The CBR-AL model achieves impressive results with values of 0.9975 for cooling (Y1) and 0.9883 for heating (Y2), indicating a high level of accuracy. The model’s success in reducing energy consumption and improving sustainability has potential ripple effects, including substantial cost savings, reduced carbon footprints, and improved operational efficiency in green buildings. This approach not only enhances environmental sustainability but also sets a benchmark for future advancements in predictive modelling for energy management.