Forecasting the future electrical load of a single apartment, a grid, an area, or even an entire country is known as load forecasting, which aims to predict future load demand. Using residential data for model training and a School-Based Optimization approach for optimising the process and computing energy consumption and occupant comfort, the proposed approach has 3 components: (1) machine learning model for low energy consumption; (2) occupant behaviour models; and (3) occupant comfort models. The experimental findings indicated that behavioural energy savings were possible, with occupant comfort significantly increased. Machine learning (ML) methods have recently contributed very well in the advancement of the prediction models used for energy consumption. AdaBoost models highly improve the accuracy, robustness, and precision and the generalization ability of the conventional forecasting which is utilized in models.