Characterization factors for global greenhouse gases emissions applying Machine Learning

Characterizing the global greenhouse gases emissions using machine learning

Abstract

Considering the promotion of sustainability practices and policies in the current economic scenario, this work present an evaluation of global socio-economic factors along with greenhouse gases emission profile based on a machine learning approach. The Random Forest algorithm was applied to perform a regression on a database with information on economic growth and greenhouse gas emissions between 1990 and 2018. The database was composed of two main groups representing the major and minor global economies selected by their economic activity level in the period. Initially, the most important features of the study were identified by means of recursive features elimination (RFE) and then the model was trained using cross validation technique before being tested. The model presented good performance metrics without overfitting. In this way, we could identify that the global greenhouse gas emission profile is impacted by specific features depending on the economic activity level. Thus, the analysis presented in this work may contribute to global governments the better understand the priorities and allocate resources wisely to promote sustainable growth.