Data-driven power consumption assessment in educational institutions using machine learning techniques

Abstract This article presents a comprehensive analysis of electrical power consumption in the educational institutions, with a focus on developing a data-driven framework for efficient energy monitoring and forecasting. As blocks accommodate classrooms, laboratories, administrative sections, and common utilities, their electrical load comprises fans, lamps, projectors, computers, laboratory instruments, and various auxiliary devices whose operating schedules vary throughout the academic cycle. Detailed data was gathered on the total number of loads, their ratings, and actual usage hours to construct an accurate representation of daily and monthly consumption patterns. To analyse and predict the energy behaviour of these blocks, Random Forest Regression a robust machine-learning method known for its superior performance in handling heterogeneous, non-linear, and noisy datasets have been selected as the primary analytical tool. The model demonstrated strong capability in capturing complex load variations and forecasting short-term and long-term power demand with high accuracy. The article uses this analysis to find peak load times, seasonal changes, and possible inefficiencies that could come from old equipment or inconsistent usage patterns. The model’s insights make it possible to suggest a number of ways to save energy, such as better scheduling of high-power devices, replacing less efficient lighting fixtures, making better use of projectors, and better load distribution across floors and time slots. The findings of this article facilitate informed decision-making regarding institutional energy planning, budget allocation, and initiatives aimed at enhancing sustainability. By combining traditional load assessment with machine-learning-based forecasting, this research creates a useful and scalable framework for smart energy management in schools. In the future, the same method could be used on all campus blocks with automated metering systems and cloud-based analytical platforms.