A Machine Learning Approach to Finding Factors that Lead to Environmental Friendliness

Abstract

                To maintain a sustainable society, environmental friendliness is necessary, an effort that all countries must take part in. The effort must be pioneered by developed nations with the resources to enact sustainable policies, reduce emissions and conserve energy, from which developing nations will follow the eroded path. Recognizing the factors that promote environmental friendliness is necessary for researchers, policymakers, and activists alike.

                Several past studies have examined the relationship between environmental performance and various nationwide factors such as economic strength, education, and corruption. In this paper, however, we introduce the machine learning approach Multiple-Linear Regression, allowing several variables to be used in tandem.

We constructed a dataset using a variety of variables from a variety of sources, either examined in past literature or justified logically. We measured environmental friendliness through the Environmental Performance Index (EPI), and chose feature variables of Women in Parliament (%), Internet users (%), Freedom Index, Ethnic fractionalization, Technological development, Press Freedom Index, Corruption Perceptions Index, GDP per capita ($), and Education Index, and Population.

                We found that Multiple-Linear Regression is an effective way of measuring EPI, where several metrics indicate that EPI is almost completely determined by the feature variables. We end the study by presenting the correlations of each of the variables with EPI, and find that almost all exhibit strong linear relationships. These correlations should bring light to the characteristics of environmentally friendly countries, mainly Nordic nations.