EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Using Machine Learning Clustering to Build an ESG-Based Taxonomy of Heat Vulnerability

Angelo Leogrande, Carlo Drago, Massimo Arnone, Alberto Costantiello

Climate · 03.07.2026 · 14 (7) · 138

DOI: 10.3390/cli14070138 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Climate, online erschienen am 03.07.2026. Verfasst von Angelo Leogrande und 3 weiteren Autor:innen. Die Arbeit nennt 128 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

HI35, defined as the mean annual number of days on which apparent temperature exceeds 35 °C, is introduced in this paper as a country-level heat exposure metric. Unlike vulnerability indicators, HI35 is treated as an exogenous climatic exposure variable and not as a direct measure of vulnerability or as an endogenous outcome. By combining HI35 with World Bank Environmental, Social, and Governance (ESG) indicators, this study applies unsupervised clustering algorithms to derive an exposure–vulnerability typology of countries. The Environmental, Social, and Governance pillars are analyzed separately through dedicated clustering procedures, supported by robustness checks and an additional country taxonomy based on principal component analysis and hierarchical clustering. The results identify heterogeneous country profiles in which similar levels of heat exposure coexist with different ESG-based vulnerability conditions, including environmental pressures, fragility, institutional capacity, innovation, water access, nutrition, and governance quality. Conversely, countries with relatively low heat exposure may display differentiated social or institutional vulnerability profiles. The empirical evidence suggests that heat exposure and ESG-based vulnerability are conceptually distinct but jointly relevant dimensions for classifying climate-risk profiles. No causal relationship is inferred between ESG indicators and HI35; the analysis is descriptive, classificatory, and based on unsupervised learning. Conceptually, HI35 captures the occurrence of extreme heat events, while ESG indicators describe the environmental, social, and institutional conditions associated with sensitivity, resilience, and adaptive capacity.

Zitieren

APA

Leogrande, A., Drago, C., Arnone, M., & Costantiello, A. (2026). Using Machine Learning Clustering to Build an ESG-Based Taxonomy of Heat Vulnerability. Climate, 14(7), 138. https://doi.org/10.3390/cli14070138

Vancouver

Leogrande A, Drago C, Arnone M, Costantiello A. Using Machine Learning Clustering to Build an ESG-Based Taxonomy of Heat Vulnerability. Climate. 2026;14(7):138. doi:10.3390/cli14070138

BibTeX
@article{Leogrande2026,
  author = {Leogrande, Angelo and Drago, Carlo and Arnone, Massimo and Costantiello, Alberto},
  title = {Using Machine Learning Clustering to Build an ESG-Based Taxonomy of Heat Vulnerability},
  journal = {Climate},
  year = {2026},
  volume = {14},
  number = {7},
  pages = {138},
  doi = {10.3390/cli14070138}
}
Kostenlos starten →