Investigating the Energy Consumption in Malaysian Educational Buildings: A Linear Regression Analysis for Baseline Energy Modeling

Abstract

The baseline energy models currently developed for Universiti Tun Hussein Onn Malaysia (UTHM) using conventional linear regression fail to meet the International Performance Measurement and Verification Protocol (IPMVP) threshold of R²>0.75, indicating insufficient explanatory power for reliable energy performance verification. The consumption of energy within the Malaysian universities is increasingly significant due to extensive facilities and growing student populations. This study analysed energy consumption patterns at UTHM and evaluated energy models using linear regression. Three independent variables included in the analysis were the number of students, the number of working days, and cooling degree days (CDD), with monthly consumption from 2022 to 2024 analysed. Simple Linear Regression (SLR) results indicated weak relationships, with correlation coefficients (R) ranging from 0.05 to 0.40. The strongest explanatory power was provided by the number of working days. The Multiple Linear Regression (MLR) enhanced the model performance, yielding coefficients of determination (R²) between 0.25 and 0.43, explaining 25%–43% of the variance. However, these values remain below the IPMVP requirement. The findings highlight the limitations of traditional linear regression and suggest the need to incorporate additional operational variables and advanced predictive approaches, such as Artificial Neural Networks (ANN) and Long Short-Term Memory (LSTM) models, to improve prediction accuracy and institutional energy planning. These findings provide useful guidance for improving energy management and conservation practices in educational institutions.