The increasing greenhouse gas emissions are the major contributors to global warming and climate change. CO \(_2\) is a greenhouse gas, and effective reduction of CO \(_2\) emission is required for sustainable development. To decrease carbon emissions and achieve several sustainable development goals, efficient and accurate forecasting is necessary before forming any preventive measures. The proposed study examines the current state of emissions and its related variables. For this purpose, the proposed FIDP method is applied to create and preprocess the dataset. After investigating various properties of the dataset and relations between variables, six machine learning models are evaluated using the cross-validation technique. The proposed ensemble model predicts carbon emissions with the highest R \(^2\) score of 97.92% and MAE, MSE, and RMSE values of near zero. The lower standard deviations across folds indicate the robustness of the proposed model. The result of this study paves the way for further optimizing emission preventive measures.