Research On Consumer Confidence Index Prediction Model Based On Differential Grey Wolf Algorithm

With the rapid development of human society and the continuous advancement of science and technology,big data as a carrier of information has increasingly revealed its pivotal position.Today's society has fully entered the information age,and the big data economy as the leading industry in the new century will become the driving force for economic development.In the face of massive data information with high randomness,contingency and non-linearity,it is very important to effectively filter data and mine information that is helpful for economic development and social progress.Therefore,more and more researchers at home and abroad are focusing on data mining and forecasting in the era of big data to increase economic returns and increase production efficiency.In recent years,domestic and foreign scholars have conducted a lot of research on time series prediction methods and models,and the prediction level has been greatly improved.Studies have shown that statistically based models are relatively simple to operate,but most of the time series in real life show strong nonlinearity.For nonlinear data columns,statistical method-based models are difficult to accurately extract valid information of data.Therefore,it is difficult to achieve a satisfactory prediction effect.The artificial intelligence neural network is used as the predictive model.The neural network method has strong learning ability,can fully extract the effective information in the data column,and solves many problems that cannot be solved by the statistical method model.However,artificial intelligence neural networks also have the drawbacks of the model itself,such as over-fitting phenomenon,large differences in the results of different prediction models etc.,making the neural network prediction model greatly reduced in time series prediction.Aiming at the above problems,the combined forecasting model based on different combinations not only fully integrates the advantages of different models,but also significantly improves the prediction accuracy and prediction effectiveness of the model.The prediction effect is often better than the prediction of a single model.Considering the above problems,this paper proposes a CEEMD-DEGWO combination model based on CEEMD decomposition denoising technology,improved DEGWO optimization algorithm and four single neural networks(BPNN,ENN,ELM,SVM),using improved differential optimization.The grey wolf algorithm optimizes the weights and thresholds of a single neural network in the combined model to construct a combined model for CCI prediction.At the same time,in order to verify the newly proposed CEEMD-DEGWO combination model has the best prediction validity,multiple rolling prediction mechanism,hypothesis testing,four model evaluation indicators,five different simulation experiments and 15 comparative prediction models are all introduced in the text.A scientific and reasonable evaluation system makes a systematic and objective evaluation of the hybrid prediction model CEEMD-DEGWO combined model proposed in this paper,and digs into the optimization ability of the algorithm and the prediction performance of the model.In order to verify the effectiveness of the proposed CEEMD-DEGWO combination model,the paper selects the consumer confidence index data from the China Economic Statistics Network Database from January 1999 to April 2018 for a total of 232 months of CCI year-on-year data.The results of the simulation experiments in Chapter 5 can be shown:15 comparison models(BPNN,DEGWO-BPNN,CEEMD-BPNN,Elman(ENN),CEEMD-ENN,DEGWO-ENN,ELM,CEEMD-ELM,DEGWO-ELM,SVM,The MAPE values of CEEMD-SVM,DEGWO-SVM,GM,equal weight combination model and multiple regression combination model are 3.47%,2.50%,2.37%,2.56%,2.16%,2.04%,2.58%,1.68%,2.08,respectively.%,3.63%,1.59%,3.10%,3.46%,1.95%,1.78%,the newly proposed CEEMD-DEGWO combination model has a MAPE value of 1.55%.Therefore,compared with the other 15 comparison models,the combined model proposed by this study has the best prediction accuracy when predicting the consumer confidence index.At the same time,the proposed model has the smallest MAPE value fluctuation at each point,which fully shows that the model can improve the accuracy and stability of CCI prediction.In addition,the discussion of DM testing and predictive availability increases the superiority of the combined model compared to other models covered herein.The consumer confidence index prediction model proposed in this paper is more advanced in the use of DEGWO optimized combined neural network for CCI prediction,and has certain forward-looking and practical significance in practical applications.The main contributions and innovations are as follows:First,based on the decomposition integration strategy,data preprocessing techniques are used to eliminate the negative effects caused by high frequency noise and extract the main features of the data.The original CCI sequence is decomposed and reconstructed into a new time series,which reduces the irregularity of CCI data and effectively improves the CCI prediction performance.Secondly,based on the swarm intelligence evolution technology,the differential evolution strategy is added to the grey wolf algorithm to realize the differential perturbation of the gray wolf individual to update the gray wolfs leadership level.The novel weighted determination method is used to optimize the combined prediction neural network model.Weights and thresholds.Thirdly,when predicting CCI,it breaks the limitations of previous studies using a single model,and innovatively uses the newly constructed combined forecasting model to effectively improve prediction accuracy and stability.Fourth,this study makes a scientific and comprehensive evaluation of the predictive performance of the combined model.The evaluation system uses four performance indicators.By comparing the size of the four indicators of different models,it is verified that the newly proposed CCI model has the best prediction accuracy.