Carbon emission is a major issue for many economics. While some have witnessed a substantial reduction in its emission, others are a warehouse of massive carbon emission. Over the years machine learning techniques have been used in predicting and visualising the impact of different economic sectors on carbon emission. Using previously published researched papers, this report seeks to use a time-series regression model – long short-term model (LSTM) for a more accurate carbon emission predictions with a 5 percent mean absolute percentage error as our baseline. Also the study shows how to develop a machine learning and Al system that will determine patterns and trends in carbon emissions over time across countries. To understand the correlation between socio economic activities and carbon dioxide emissions using graphic user interface and other visualization tools, and to predict global carbon emissions based on the socio-economic activities of 66 countries.
CO₂ emission intensities are calculated by dividing the CO₂ emissions from fuel consumption by output from the OECD Inter-Country Input-Output (ICIO) Tables and multiplying the result by 1 million for scaling purposes.
CO₂ emission multipliers are calculated by multiplying the Leontief inverse (also known as output multipliers matrix) from the OECD Inter-Country Input-Output (ICIO) Tables by the CO₂ emission intensities.