AUTHOR=Kong Depeng , Dai Zheng , Tang Jiayue , Zhang Hong 

TITLE=Forecasting urban carbon emissions using an Adaboost-STIRPAT model

JOURNAL=Frontiers in Environmental Science

VOLUME=Volume 11 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/environmental-science/articles/10.3389/fenvs.2023.1284028

DOI=10.3389/fenvs.2023.1284028

ISSN=2296-665X

ABSTRACT=Solving outstanding environmental issues, reducing carbon emissions, and promoting green development are necessary ways to achieve carbon neutrality and carbon peak goals. It is also an important issue faced by society today. This paper uses the Kaya identity combined with the logarithmic mean Divisia index (LMDI) decomposition method to analyze the factors affecting carbon emissions, and uses the Pearson correlation coefficient to screen out eight highly correlated features to construct an extended STIRPAT model. In order to further improve the accuracy of the model in predicting carbon emissions, this paper introduces the Adaboost algorithm from machine learning to enhance the STIRPAT model. Finally, scenario analysis is used to predict and analyze carbon emissions in Shandong Province from 2020 to 2050. The results show that: (1) The main factors affecting urban carbon emissions from 1998 to 2019 are economic growth effects, followed by energy structure effects and energy consumption effects. (2) Under three different development scenarios, Shandong Province can achieve carbon peak between 2030-2035, but there are differences in peaking time and peak values. This is a provisional file, not the final typeset article recent times (Hepburn et al., 2021; Jones et al., 2008). Research in this realm primarily diverges into exploring the factors driving carbon emissions and prognosticating future emissions trajectories (Steenhof, 2007;Whitmarsh et al., 2010). This study aims to fill the existing knowledge voids, particularly by amalgamating advanced machine learning algorithms to improve the accuracy of predictive models.The body of literature on carbon emissions is extensive and diverse. For dissecting the factors influencing emissions, the amalgamation of the Kaya identity and the Logarithmic Mean Divisia Index (LMDI) decomposition method has proven instrumental, shedding light on key drivers such as energy consumption, economic indicators, and demographic transitions (Ang & Liu, 2001;Wang et al., 2005). Numerous other methodologies, from structural decomposition analysis to gray correlation degree models, have been employed, unearthing fundamental determinants like economic scale,