Prediction Of The Prosperity Index Of Macro-economic Based On CEEMD And Wavelet Analysis

The macro economy can reflect the overall economic situation of a country or region.The trend of macroeconomic development not only affects the government’s macroeconomic policies,but also affects the expectations of corporate and individual investment decisions.Therefore,accurately predicting the development trend of macroeconomics has important practical significance.Based on the machine learning algorithm,this paper establishes a combined model to predict the macroeconomic prosperity index.In this paper,a combined model of neural network method and time series decomposition is constructed for prediction.The article takes China’s macroeconomic prosperity index as the research object.Joint denoising of the original time series of economic prosperity index by CEEMD and wavelet decomposition is used.And then,the machine learning algorithm is used for prediction.First,the time series is decomposed by CEEMD to obtain multiple IMF components and residual waves,and then the high frequency components are decomposed by wavelet decomposition.Data reconstruction is performed on the denoised IMF and the residual wave.Then SVM,BP neural network are used for prediction,and cross-validation,particle swarm optimization algorithm are introduced to improve prediction accuracy.The results show that the proposed PSO-BP model based on joint denoising and SVM model is better than CEEMD denoising model,wavelet denoising model,and the original model without denoising.The neural network model based on CEEMD-wavelet decomposition combined denoising can be applied to predict of macroeconomic prosperity index(consistent index),which provides an important reference for macroeconomic regulation,corporate and individual investment decisions.