Carbon Price Analysis And Forecasting Based On Independent Component Analysis

In the past few years,along with the stronger and stronger development of carbon market,the carbon price fluctuation has drawn the attention of researchers,encouraging numerous researchers involved in the carbon price study.However,ow-ing to the strongly nonlinear and nonstationary characteristics of carbon price,most of existed approaches failed to forecast the carbon price perfectly.In this paper,a novel hybrid forecasting model based on variational mode decomposition(VMD),independent component analysis(ICA)and radial basis function neural network(RBFNN)is presented to analyze and forecast the carbon price.In this model,firstly,VMD is utilized to decompose the carbon price data chosen into several in-trinsic mode functions(IMFs).Next,the IMFs arc regrouped by the correlation analysis method into a new set of series called NIMFs.Then ICA is applied to the NIMFs in order to extract the independent components(ICs)and corresponding practical factors are endowed to them,including CER price,market economy,his-torical events,climate change and so on.Thercout we can know what the influence factors are and how they work on the fluctuation of carbon price.After that,the au-thors give a description for the availability of the results above and the feasibility of the presented model.Finally;the ICs are trained by RBFXN to predict them respec-tively and the final forecasting result of the original carbon price data is obtained by linear combination of each independent component prediction result.Moreover,the numerical experiments are performed to verify the effectiveness of proposed hybrid forecasting model.The numerical results show that the proposed model is able to outperform the single ARIMA,RBFNN models and combing VMD-RBFNN,W-NN and EMD-ICA-RBFNX models:indicating the accuracy and feasibility of our proposed hybrid VMD-ICA-RBFNN model.This paper contains five chapters,the details are illustrated as following:Chapter 1 Introduce the three basic methods VMD,ICA and RBFXN in the proposed VMD-ICA-RBFNN model.Chapter 2 Introduce the hybrid VMD-ICA-RBFNN model,indicating the merit of the model and give the chief steps of the model.Chapter 3 Use VMD and ICA to decompose the carbon price data series,obtaining a series of ICs.And give each of IC a statistic description,while corre-sponding practical factors are endowed to them.We analyze how the factors affect the fluctuation of carbon price and do the work of feasibility analysis.Chapter 4 Based on the results in the Chapter 3,we apply the RBFNN technique on each of IC respectively,and regroup the forecasting results according to the theory of VMD and ICA to get the final forecasting resultChapter 5 For certifying the forecasting effect of the proposed VMD-ICA-RBFNN model better,the authors utilize the single forecasting models ARIMA,RBFNN and hybrid models VMD-RBFNN,W-NX-and EMD-ICA-RBFNN models to have a comparison with the proposed model,at the same timc we adopt some evaluation indexes to show the high precision of our model.