A novel data-driven rule-base approach with driving factor decomposition for multi-scenario prediction on carbon emission reduction

Abstract

Reducing carbon emissions is an ongoing goal of the whole world and its achievement requires an outstanding approach to accurately predict future carbon emissions and explore the factors driving carbon emissions. Hence, this study proposes a driving factor decomposition-based data-driven rule-base (DFD-DDRB) approach for the aim of analyzing carbon emission reduction pathway from predictive perspective, where the approach includes three processes: 1) generating a rule-base from historical carbon emission data; 2) predicting multi-scenario carbon emissions using the rule-base; 3) providing predictive analytics for future carbon emission reduction. In empirical study, the China’s provincial data from 2004 to 2021 are used to justify the applicability of the proposed approach. The experimental findings not only show that the approach can accurately predict multi-scenario carbon emissions until 2035 and reveal the factors driving carbon emissions, but also provide three implications for reducing China’s carbon emissions: 1) resource endowment should be considered to establish carbon emission management policies of 30 Chinese provinces; 2) economic development effect can be regarded as the main factor driving China’s future carbon emissions; 3) optimizing energy structure and consumption is much important for reducing China’s provincial carbon emissions. Beside the work in China, the DFD-DDRB approach can be also used as the generic analytical framework served for some developed economies and other carbon-emitting countries.