A novel ensemble approach for road traffic carbon emission prediction: a case in Canada

The "Annual Report 2021" from the United Nations Environment Programme (UNEP) highlights that the transportation sector is the fastest-growing greenhouse gas emissions sector, accounting for approximately 25% of energy-related emissions. What is even more concerning is that, at a time when carbon emissions need to be urgently reduced across various industries globally, carbon emissions from the transportation sector continue to rise. This is because the improvement in the efficiency of vehicle power combustion struggles to offset the increasing emissions resulting from the massive volume of travel. With the enhancement of transportation networks in various countries, it is projected that the growth rate of carbon emissions in the transportation sector will surpass that of the industrial and power sectors, presenting a significant challenge to achieving the emission reduction goals outlined in the Paris Agreement. Carbon emissions in the global transportation sector encompass various modes of transportation, including road, rail, aviation, and maritime, with road transportation being the largest contributor to carbon emissions. This study utilized the Stacking technique to build the X-MARL model for predicting $$\hbox {CO}_{2}$$ CO 2 emissions from vehicles and formulated recommendations for carbon reduction in the transportation industry. The model was tested using a dataset of vehicle $$\hbox {CO}_{2}$$ CO 2 emissions officially recorded by the Canadian government, comprising 7385 data points and covering 12 different vehicle parameter attributes. During the experimentation process, three statistical evaluation metrics were employed, namely mean squared error (MSE), root-mean-squared error (RMSE), and the coefficient of determination (R2). The dataset was randomly split into a training set (80% of the total data) and a testing set (20% of the total data). The experimental results demonstrated that the X-MARL model exhibited the highest prediction accuracy. This study provides an original strategy for accurately predicting carbon emissions from road transportation, which can offer support and guidance to decision-makers in formulating and implementing effective environmental policies.