Published on July 2026 | Machine Learning, Deep Learning, Artificial Intelligence

AI-DRIVEN PREDICTION OF BUILDING ENERGY AND COMFORT

Since the building sector accounts for around 40% of global energy consumption and 33% of greenhouse gas emissions, energy-efficient building design is an essential objective for sustainable development. This paper presents the Climate AI Design Predictor, an end-to-end, machine-learning-powered framework that uses climate conditions and architectural design parameters to forecast building energy consumption (heating load, cooling load, and total energy consumption) and the thermal comfort index. Meteorological data from five Indian cities—Bangalore, Chennai, Delhi, Jaipur, and Shimla—obtained from NASA’s POWER API is combined with the UCI Energy Efficiency dataset. Two ensemble learning algorithms, Random Forest (RF) and Light Gradient Boosting Machine (LightGBM), are trained and compared. Experimental results demonstrate exceptional predictive performance, with the best models achieving R2 scores of 0.9972, 0.9680, 0.9926, and 0.9998