Uncovering the significant socio-economic attributes of low- and high-emission countries using rough sets

Abstract

Drivers to global carbon emissions have been widely investigated in the scientific literature. However, most previous studies have been limited to supervised learning approaches such as decomposition analysis. Thus, the effects of more specific socio-economic factors such as research expenditure, poverty incidence, education level and trading of goods have seldom been probed into, and often, drivers significantly vary from country to country. However, it is hypothesized in this study that patterns can be derived among high- and low-emission countries using a more detailed approach. Thus, a novel approach using rough sets is developed to uncover the effects of detailed socio-economic attributes to the emissions of 194 countries and regions. A significant advantage of rough set theory is its ability to work with incomplete data sets. As comprehensive as they are, global data sets such as that of the World Bank would still have gaps especially in less developed countries. The rough set model developed in this study has a validity of 94%. The most significant factors for low-income countries are having low to mid agricultural exports and having mid to high pump prices for diesel. For high-income countries, this was having high gross domestic product per capita. Overall, the most interesting insight from this study is that countries do not simply grow as they increase emissions, but also evolve. As countries develop, they also change their priority sectors, policies and demographics. Moreover, the findings suggest that the understanding of low-emission countries benefits more from a comprehensive study. High-emission countries have been well studied in the literature already.