Smart Energy Assessment in Educational Institutions Using Machine Learning Approaches

Objective: To develop a data-driven framework for accurately assessing, forecasting, and optimizing electricity consumption in educational institutions to support efficient energy management, cost reduction, and campus-scale sustainability planning. Method: detailed inventories of electrical loads were prepared for B and C blocks of Methodist College of Engineering and Technology, Hyderabad, India in the academic year 2024-25 by documenting appliance types, rated capacities, quantities, locations, and operating schedules, which were used to model daily and monthly demand patterns; to handle heterogeneous and non-linear consumption behaviour driven by occupancy, academic timetables, equipment diversity, and seasonal effects, a Random Forest Regression model was trained on historical measurements and evaluated using structured validation procedures, including cross-validation and standard statistical error metrics, to support both short-term load forecasting for operations and longer-term demand estimation for capacity planning. Findings: the proposed framework successfully characterized consumption behaviour across classrooms, laboratories, offices, and shared facilities, revealing