The application of machine learning in the development of carbon-neutral cities: a systematic literature review

Addressing climate change requires cities to transition toward carbon neutrality. In this context, machine learning (ML) has emerged as a crucial technology that supports data-driven decision-making processes to facilitate and accelerate this transition. Even if the number of studies in this sector has significantly increased, there is still a lack of a thorough and organized panorama of the state of the research. In order to map the current state, trends, gaps, and future directions in the application of machine learning for the construction of carbon-neutral cities, this work presents a Systematic Literature Review (SLR) in accordance with PRISMA principles. The results of this study, which examined 54 important articles published between 2020 and 2025, show that this field of study has grown exponentially and has a notable geographic focus on China. Methodological examination shows that supervised learning paradigms are highly prevalent in this discipline, especially neural networks like Long Short-Term Memory (LSTM) and tree-based models like XGBoost and Random Forest. With half of the studies, the "Energy and Buildings" sector is the main area of application study, followed by "Urban Policy and Governance," "Transportation," and "Waste Management." The research flow is mapped from the "Heterogeneous Data Ecosystem" to the "Machine Learning Analytical Core," then to the "Integrated Application Domains," and lastly to the "Operational Outputs," according to a novel four-stage conceptual framework proposed in the article. This paradigm provides academics and policymakers with a strategic road map for achieving carbon-neutral cities by highlighting the progression of research from descriptive and predictive studies to prescriptive and automated decision-making support.