Machine Learning Insights into Nordic CO2 Emission Trends

Abstract

Rapid industrial development has substantially increased carbon emissions, leading to heightened concentrations of greenhouse gases and resultant climate change . This phenomenon poses diverse threats, including risks to food security, water availability, extreme weather events, disease proliferation, economic downturns, and population migration. Recognizing climate change as the greatest threat to global health, the World Health Organization (WHO) emphasizes its significance. Since 1970, CO2 emissions have surged by 90%, comprising 78% of total greenhouse gas emissions. Predicting these emissions is challenging due to dynamic scenarios influenced by climate impacts, carbon factors, and socio-economic attributes, rendering accurate prediction crucial yet complex. To address this complexity, artificial intelligence and machine learning techniques are increasingly utilized to study environmental phenomena characterized by high variability. This paper conducts a thorough analysis of CO2 emission predictions for Nordic countries (Denmark, Norway, Sweden, Finland, and Iceland) spanning from 2018 to 2029. Employing data exploration, visualization, and machine learning techniques, the study leverages the International Greenhouse Gas Emissions dataset, spanning from 1990 to 2017. The research seeks to unravel historical emission trends and forecast future CO2 levels. Methodologically, it encompasses data preprocessing, exploratory analysis, and the application of machine learning models, including multiple linear regression, ridge regression, lasso regression, and polynomial regression. The findings reveal varying predictive capabilities, with polynomial regression emerging as the standout performer. In the context of model performance, the polynomial regression model exhibits noteworthy results, with a mean absolute error (MAE) of 18667.84, a root mean squared error (RMSE) of 53277.75, an R-squared value of 0.995, and an explained variance of 0.995. This superior performance positions polynomial regression as a robust choice for predicting CO2 emissions, particularly in capturing nonlinear relationships in environmental phenomena. The analysis extends further, encompassing a detailed examination of country-specific observations, yearly changes, and recommendations for effective emission reduction strategies.