Application Of Combinatoria Model Of Neural Network And ARIMA

In recent years,scientists have made extensive use of artificial neural networks,and the prediction field has been greatly developed with the introduction of neural network.GDP is an important indicator to measure a country’s development level.The country can adjust and formulate relevant policies through the situation reflected by GDP.Therefore,research on GDP prediction is very meaningful.At the same time,consumer price index(CPI)is one of the important indicators to analyze the basic trend of China’s market price,and it is also an important basis for national macro-control.The CPI prediction is also very important.In this article,we believe that a single neural network model is limited in mining linear and nonlinear information contained in sequences.In order to further improve the predictive effect of these two types of data,we compare neural networks with conventional time series.Model ARIMA model.Combine building the mixed model and then apply the mixed model to GDP and CPI forecasts.The main research contents of this thesis are as follows:This article briefly presents the relevant knowledge about neural networks and the theory of time series and hybrid models.This lays the theoretical basis for the full text.First,a single neural network model was built to predict GDP and CPI data,then an ARIMA model was built to predict both,and finally a neural network and ARIMA hybrid model with BIP were built.Applied to CPI.Compare and analyze the predictive effect of a mixed model of two types of data prediction with the predictive effect of a single neural network model.