Energy consumption prediction in educational building during lecture week using multiple regression model / Rijalul Fahmi Mustapa … [et al.]

Energy consumption prediction in educational building is vital before any activities pertaining to energy management (EM) and energy conservation measures (ECM) were conducted. The successfulness of the mentioned activities relies on the accuracy of energy consumption prediction through a baseline energy model (BEM). BEM is a tool that were used to assess the changes and behaviour of energy consumption in any buildings. Currently, Single Linear Regression (SLR) model were widely used for BEM modelling and prediction purposes. SLR model that were used for BEM modelling and prediction purpose poses significant weakness where it relies only on one independent variable as the contributing factor for the energy consumption. Thus, the innovation objective of this project is to use the Multiple Linear Regression (MLR) model for BEM modelling and energy consumption prediction purposes in an educational building. An educational building will be used as a case study where SLR model and MLR model will be used for BEM modelling hence predicting the energy consumption. Both model will be compared using coefficient of determination (R2) and statistical error MSE, RMSE and MAPE to assess its reliability and prediction accuracy respectively. Results demonstrate that MLR model has a higher R2 and low statistical error compared to SLR model which concludes MLR model has higher advantages and accuracy in modelling and prediction. In addition, it is safe for building owner if they intend to use the MLR model for BEM modelling and energy consumption prediction purposes.