Forecasting the Carbon Emissions in African Countries Using EMD-GCN

Different countries in Africa region are recognized as the top energy consumers and CO2 emitters. With South Africa significant harm to living organisms and more emissions of 7.34 metric tons per capita, this paper aims to address the detrimental effects of CO2 emission in top African countries. The study focuses on predicting CO2 emission for the next 10 years, utilizing timeseries data from 1980 to 2021. To achieve this, two machine learning models (GCN and EMD-GCN) were employed. This paper presents a comprehensive approach by incorporating a range of models for data prediction. Performance analysis reveals that among the 2 models, graph based method with series decomposition EMD demonstrate the highest accuracy based on 4 performance metrics. A comparative study further corroborates these findings. Consequently, the deep learning-based GCN model is recommended as one of the most suitable models for accurate CO2 emission prediction.