Achieving China's CO2 reduction targets: Insights from a hybrid PPA-PPR forecasting model

Abstract

China is the largest carbon dioxide (CO₂) emitter and has formulated CO₂ emission peak and carbon-neutral plans. Studies on CO₂ emission volume and CO₂ emission intensity (CEI) indicate a growing interest in related fields. The purpose of this research is to improve the performance and reliability of the model for forecasting CO₂ emissions and judging whether to achieve China's CO₂ reduction targets under business-as-usual scenarios. We originally applied a novel hybrid model combining the projection pursuit regression (PPR) model and parasitism-predation optimization algorithm (PPA) (PPA-PPR) to forecast the CO2 emissions from 2022 to 2035, with the time series' CO2 emissions data from 1965 to 2017 and compare with various machine learnings. From the studied results, we can conclude that the hybrid PPA-PPR has better stability and accuracy than BPNN, RF, SVM, GM, TDGM, and LSTM models. The mean absolute percentage error (MAPE) of the verification set data is only 1.19%, and the MAPE of forecasting set data from 2018 to 2021 is 1.44, which obviously outperforms the BPNN, RF, SVM, GM, TDGM, and LSTM models. The second finding is that if China continues to develop in its present trend, it can't implement the CO₂ emission peak and reduction CEI target by over 65% by 2030. The limitation of this research is that we don't decompose the time series data into intrinsic modes to study the possibility of improving the model performances.