Uncovering Global Emission Regimes Using Unsupervised Learning: A Structural and Dynamic Analysis

Abstract

Understanding the structural composition of greenhouse gas (GHG) emissions is critical for effective climate mitigation, yet existing research largely focuses on aggregate levels rather than the multivariate structure of sectoral emissions. This study develops a data-driven framework to identify emission regimes using a global panel dataset for 172 countries over 1990–2023. Sectoral emissions are transformed into compositional vectors to capture relative contributions, followed by Principal Component Analysis (PCA) and K-means clustering to uncover latent structural regimes. The first two principal components explain 37.9% of total variance and reveal clear gradients distinguishing agriculture-dominated versus industrial-energy systems and fossil extraction versus urban-consumption emissions. Five distinct regimes are identified based on silhouette optimization (s̄ = 0.38). A first-order Markov transition analysis shows strong persistence, with diagonal probabilities ranging from 0.876 to 0.973, indicating high path dependence and limited short-term mobility across regimes. However, gradual transitions toward mixed emission systems suggest slow structural diversification. Robustness checks confirm stability, with between-cluster variance explaining 68.5% of total variation. Overall, the findings show that global emission systems are not homogeneous; rather, they exhibit distinct structural patterns across regions, strong persistence over time, and only gradual transitions between regimes. These findings highlight the importance of moving beyond aggregate emission trends and focusing on structural emission regimes to better understand the long-run evolution of global greenhouse gas emissions.

Declarations

Consent to publish: The authors have provided consent to publish.

Competing interests: The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Data availability statement: The data used in this study are available from publicly accessible sources.

Funding: The authors received no financial support for the research, authorship and/or publication of this article.

Clinical Trial Number: Not Applicable.

Declaration of using generative AI: During the preparation of this work the author(s) used ChatGPT in order to correct the grammatical errors. After using this tool/service, the author(s) reviewed and edited the content as needed and take full responsibility for the content of the published article.

Author Contributions: R.E.A.R.: Conceptualization, methodology, writing—original draft preparation, writing—review and editing. All authors have read and agreed to the published version of the manuscript.