Research On Prediction Of Stock Market Index Based On CEEMDAN-FE-NAR

Financial time series has the characteristics of high noise,instability and nonlinearity,which are contrary to the assumptions of traditional measurement models,so the traditional measurement models can not deal with financial time series well.In recent years,the rise of computer technology has promoted the rapid development of time-frequency domain analysis methods and machine learning.Compared with traditional statistical analysis methods,time series prediction methods such as fully adaptive ensemble empirical mode decomposition(CEEMAN)and NAR neural network show strong processing ability and application adaptability in dealing with nonlinear and non-stationary data,and have broad prospects in the field of finance.At the same time,as the most important stock market index in China,Shanghai stock index is an important indicator signal of China’s economic and social development.It is of great significance to predict the changes of Shanghai stock index combined with CEEMAN,NAR neural network and fuzzy entropy(FE)algorithm.This paper first summarizes the relevant theoretical research results,then introduces the relevant theories of CEEMAN,in detail,and expounds the principles of Fe algorithm,NAR neural network and ARIMA model,so as to provide theoretical basis and method preparation for predicting Shanghai stock index.On this basis,the Shanghai stock index from February 1,2016 to July 21,2020 is taken as the research object.CEEMAN is used to decompose the Shanghai stock index into 10 eigenmode functions(IMF)with different fluctuation frequencies and a residual term,and Fe algorithm is used to calculate the fuzzy entropy of each IMF.Then,according to the calculated fuzzy entropy,the IMF is reconstructed into highfrequency component,medium frequency component and low-frequency component,and then the high-frequency component The periodic analysis and correlation analysis of intermediate frequency component and low frequency component reveal the fluctuation characteristics of different components and the internal structure component of Shanghai stock index fluctuation.Finally,based on the reconstructed components,ARIMA model and NAR neural network are used to predict the three components respectively,and then the predicted values of each sequence are obtained,and the final prediction results are obtained by summing.By comparing the first mock exam results with other models,such as EMD and EEMD,the following conclusions can be drawn:(1)compared with the single model,the CEEMDAN-FE-NAR composite model has higher accuracy in forecasting Shanghai Composite Index.(2)CEEMAN-FE-NAR neural network has the best prediction ability;(3)CEEMAN decomposition method can improve the prediction ability of the prediction model better than EMD and EEMD.