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.

Cluster and regression analyses to model global emissions