As climate change continues to escalate, the call to action for countries’ carbon emissions reductions is paramount to global sustainability. This study explores, in the context of the Sustainable Development Goals (SDGs), the extent of emissions generated and the emissions mitigated from green innovation, green finance, green taxation, and green accounting. We used a machine learning approach in which we implemented Extreme Gradient Boosting Regression (XGBoost) to explore relationships that are complex and non-linear behaviour of these variables relative to CO₂ emissions. XGBoost offers performance improvements over traditional regression models; moreover, it generates feature importance scores that clarify the strongest policy levers for change. For sustainability, XGBoost leverages the heterogeneous aspects of the environmental and economic data, as the data are collected across different policies and geographies. The dataset was generated by the World Bank and underwent preprocessing, exploratory data analysis, and model validation to verify the quality of the data and reliability of the model selection. Our results revealed that green taxation serves as the strongest contribution to emissions reductions, followed by green finance and innovation. The feature ranking was also useful as it brought some transparency to the decision-making process within the model, which is also an important aspect of machine learning and policy research. These results provide an evidence base to modify sustainable development plans and highlight the role of AI in environmental policy modelling, particularly regarding SDG 7 (Clean Energy), SDG 9 (Industry and Innovation), and SDG 13 (Climate Action).
Keywords: Carbon emissions, climate change mitigation, sustainable development, green innovation, green finance, carbon pricing, XGBoost regression analysis, emission quantiles, Sustainable Development Goals (SDGs), emission reduction, green policies, global sustainability, policy measures.
