In an era of rapid urbanization and environmental degradation, sustainable urban development is imperative to ensure a high quality of life while reducing carbon footprints. This paper presents a comprehensive framework that leverages artificial intelligence to drive data-informed decisions in smart cities, ultimately aiming to create sustainable and carbon-free urban environments. Our approach integrates diverse datasets representative of key urban challenges including energy efficiency, air quality, infrastructure durability, and both residential and industrial energy consumption into a unified predictive modeling platform. Utilizing PyCaret's low-code machine learning (ML) library, we automated the training, selection, and evaluation of numerous regression models, with a particular focus on ensemble-based methods such as Extra Trees, CatBoost, and LightGBM. Rigorous benchmarking across six publicly available datasets demonstrated near-perfect predictive performance, with R2 values often exceeding 0.99 and minimal error metrics observed in multiple domains. These results highlight the models' robustness and suitability for high-stakes applications in urban sustainability, ranging from energy optimization to environmental monitoring. The performance benchmarking presented in this paper serves as a practical validation of the proposed AI-driven framework, covering essential smart city domains such as environmental monitoring, infrastructure resilience, and energy efficiency. The study not only underscores the potential of AI in transforming urban infrastructure but also provides a scalable and interpretable framework for real-world deployment. By converting vast, heterogeneous urban data into actionable insights, our work paves the way for smarter, carbon-neutral cities, where predictive analytics serve as the cornerstone of sustainable urban policy and operational excellence.
Salhi et al. (2025) studied this question.