Country-Level CO₂ Emissions Forecasting Using Machine Learning: An Explainable AI Analysis of the GDP Factor

In the fight against climate change, accurately predicting the future trajectory of carbon dioxide (CO2) emissions and understanding the key drivers behind these emissions is of critical importance. While the impact of economic growth—typically measured by Gross Domestic Product (GDP)—on emissions is well-documented, interpreting this influence through the lens of modern predictive models remains a significant area of inquiry. This study aims to develop a high-accuracy emission forecasting model using country-level CO2 emission data and to conduct an in-depth analysis of the influence of the GDP attribute on model outcomes by employing Explainable Artificial Intelligence (XAI) methods. The research utilizes machine learning models that incorporate time series characteristics and macroeconomic variables such as GDP. Model performance is assessed using standard evaluation metrics, including R2, MAPE, RMSE, MAE, and RRSE. To examine the influence of GDP on model decision-making, particular emphasis is placed on SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations) techniques. These tools help assess not only the overall importance of the GDP attribute but also its positive or negative contributions to individual predictions, as well as the nature of its relationship with the target variable—whether linear or nonlinear. By highlighting the predictive power of machine learning in the context of emission forecasting, this study also leverages XAI tools to move beyond the "black-box" nature of complex models. It offers a more transparent understanding of how economic factors, particularly GDP, influence climate dynamics. The findings are expected to shed light on the complex GDP-emission relationship and support the development of evidence-based climate policies.