Data-driven predictive models for sustainable smart buildings

Buildings account for about 40 percent of global energy consumption. Merging artificial intelligence with smart architecture presents numerous opportunities to transform these spaces into sustainable environments, hence evolving them into eco-friendly smart green buildings that contribute positively to the ecosystem. In intelligent buildings, AI enables the analysis of vast amount of data collected from sensors and building systems, facilitating enhanced energy management, predictive maintenance, tailored comfort controls, and operational efficiencies derived from informed decisions based on operational conditions and occupant behavior trends, which aid in maximizing both functionality and sustainability of a building. Green AI is crucial for enhancing the environmental sustainability of AI systems in intelligent buildings. While some research is being undertaken on the role of AI in green and smart architecture, our comprehension of its usability and advantages is fragmented. The integration of renewable energy sources in smart buildings fosters a synergistic relationship between clean energy generation and smart energy management systems. This not only enhances the sustainability of individual buildings but also contributes to the creation of greener urban environments and more sustainable energy frameworks. This paper delves into the potential impact of machine learning on fostering sustainability within smart buildings. It highlights the critical role of energy efficiency and the importance of lowering carbon footprints through the implementation of advanced algorithms, including K-Nearest Neighbors (KNN), Support Vector Machines (SVM), Random Forest, XGBOOST, AdaBoost, and Naive Bayes classifiers. By utilizing these sophisticated techniques, the path toward enhanced sustainability in building management becomes increasingly viable. Moreover, the paper presents ANOVA (Analysis of Variance) as a valuable methodology for feature selection. This analytical tool is instrumental in pinpointing the most influential attributes necessary for effective model training. By streamlining the feature selection process, ANOVA not only enhances the performance of the models but also conserves computational resources, making the approach more efficient overall. Through this investigation, the paper underscores the transformative potential of machine learning in achieving sustainable smart building practices.