Research On Carbon Emission Prediction In China Based On Adaboost-LASSO

In the global trend of energy conservation and emission reduction,China has actively responded to and established emission reduction targets and formulated a series of policies to achieve reasonable emission reductions.2020 is the year when China's "Thirteenth Five-Year Plan" ends,and it is also the node for China's various achievements acceptance.On the one hand,it is necessary to check the energy saving and emission reduction targets at this stage,find gaps and implement them,and improve the emission reduction path.On the other hand,it is necessary to formulate a reasonable policy plan for the next new stage in order to achieve more efficient and stable development.Precise carbon emissions forecasting is a prerequisite for policy development guidance and planning.However,because carbon emissions are affected by many factors such as energy,economics,social development,industrial structure,and technology,and different factors have different degrees of influence,how to extract valid information from many variables for a more accurate prediction is carbon emissions prediction The essential.Based on the above background,this paper proposes a prediction model based on the Adaboost lifting algorithm to optimize LASSO regression to predict and analyze China's carbon emissions.First,this article selected indicators covering many levels including energy,economy,industrial structure,social development,and scientific and technological progress,and initially classified 19 indicator variables with carbon emission sources as the classification criteria.On this basis,based on the theory of econometrics,this paper tests the influence factor variables to verify the rationality of the influence variable selection.One of the cores of this research is to use the LASSO algorithm to extract the above variables and further expand the prediction.Secondly,this paper verifies the accuracy of the prediction results and the validity of the model by performing longitudinal comparison analysis before and after the proposed model and horizontal comparison analysis with other regression methods.In addition,this paper uses the proposed model and scenario analysis to forecast and analyze China's carbon emissions in 2020,2025,and 2030,and simulates carbon emissions in different scenarios.Finally,based on the analysis of different influencing factors and the results of scenario simulation predictions,this paper proposes improvements and measures to provide an effective reference for policy formulation.The research results show that the results of index screening using LASSO show that coal consumption,oil consumption,flat glass production,and pig iron and crude steel production have a greater impact on carbon emissions.Compared with the average relative error of 0.99% predicted by the single LASSO model,the average relative error of the Adaboost-LASSO model is only 0.23%.And its error is far lower than regression models such as linear regression,ridge regression,elastic net regression,and the improved regression model.Scenario prediction results for carbon emissions show that under the baseline scenario,China's carbon emissions are predicted to be on the rise.Under a policy-driven scenario,China's carbon emissions first rose and then fell to a peak in 2018.Under the technology-driven scenario,China's carbon emissions have been on a downward trend since 2019.