Intelligent Energy Management: Machine Learning-Based Electricity Consumption Prediction for Various University Buildings

Enhancing electricity management in university buildings is essential, as it aligns with Saudi Arabia's Vision 2030 goals of sustainability and economic diversification. To address this, several machine learning-based models were developed to predict electricity consumption across various buildings at Princess Nourah bint Abdulrahman University (PNU). This study utilized 3,588 records obtained from the University Asset and Investment Agency at PNU between December 2022 and February 2024. Four regression models-Linear Regression, K-Nearest Neighbors, Gradient Boosting Regressor, and Random Forest -were designed, trained, and tested. Their performance was evaluated and compared based on Mean Squared Error, Root Mean Squared Error, Mean Absolute Percentage Error, and R-squared. The Gradient Boosting Regressor demonstrated superior performance, achieving a Mean Squared Error of 0.2 percent and an R-squared value of 85.18 percent, validating its effectiveness.