Title: Cluster and regression analyses to model global emissions

Authors: Celso da Silveira Cachola; Jhonathan Fernandes Torres de Souza

Addresses: Institute of Energy and Environment (IEE), University of São Paulo and National Industrial Training Service (SENAI), São ‎Paulo, Brazil ' Brazilian Council for Sustainable Construction (CBCS), São ‎Paulo, Brazil

Abstract: This paper analyses country clusters based on their greenhouse gas emission profiles by sector and per capita income. We used the k-means algorithm to cluster countries and tested distinct regression models to assess their statistical performances, forecasting G20 countries' per capita emissions up to 2050. Highest emissions per capita occur in cluster 5 composed by countries, e.g., Israel and Qatar. The multinomial/hierarchical model has shown better performance than linear regression according to log-likelihood. The findings in this article can provide insights for joint strategies among similar countries, and promote the use of k-means algorithm for multilevel regressions and cluster analysis.

Keywords: climate change; carbon dioxide emissions; cluster analysis; hierarchical regression.

DOI: 10.1504/IJGW.2024.142863

International Journal of Global Warming, 2024 Vol.34 No.4, pp.300 - 313

Received: 14 Apr 2024
Accepted: 07 Sep 2024
Published online: 28 Nov 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article