Study On Spatio-temporal Transmission Effect And Prediction Of Apple Price

In recent years,due to natural disasters and other impacts,the imbalance of supply and demand has caused frequent and large fluctuations in the price of the Apple market,which has brought severe challenges to the sustainable and healthy development of the Apple industry.Accurate prediction of apple prices can guide the adjustment of industrial planting structure and business strategy and is an important means to stabilize the healthy development of the apple industry.This article takes historical apple price data as the research object,analyzes the spatio-temporal transmission effect of apple price,the factors affecting apple price fluctuation,and the characteristics of the apple price series.This paper constructs and optimizes a short-term forecast combination model based on Shandong apple prices.First,analyze the temporal and spatial transmission effects of apple prices,and use the vector error correction model and impulse response function to analyze the characteristics of the regional time transmission effects of short-term apple prices and long-term apple prices respectively.It also analyzes multiple factors affecting the price of apples,including apple yield,planting area,total planting cost,and GDP,and the time transmission effect on the price.Using the Gini coefficient,Theil T index,traditional Markov chain,and spatial Markov chain,the spatial transmission characteristics of apple prices are analyzed from the perspective of regional gap and dynamic evolution.According to the analysis of short-term and long-term price transmission effects,the price of apples between provinces and cities has a time transmission effect and price transmission is asymmetric.Apple yield,planting area,total planting cost,and GDP have positive or negative effects on apple prices.The research on the regional gap and dynamic evolution of apple prices found that there is a spatial transmission effect of apple prices in various regions,and there is a relatively obvious and stable regional gap.Second,construct the GARCH family model to obtain the economic characteristics of price time series,such as volatility clustering,volatility persistence,asymmetry,and leverage effect,and then propose a combination of LSTM recurrent neural network and econometric model-new method.The analysis of the prediction results of multiple models shows that the combined model combining LSTM and GARCH family has the best prediction performance.And the combined model with the TGARCH term has the highest overall prediction accuracy,It proves that the asymmetric volatility effect of the apple price series has a greater impact on price fluctuations.At the same time,after comparing the prediction accuracy of multiple combined models,it is found that the prediction accuracy of the model is wirelessly correlated with the number of LSTM combined GARCH family models.Finally,to further improve the prediction accuracy of the apple price,optimization and improvement are made based on the combined model of the LSTM and GARCH family,which increases the information entropy of the price transmission effect in time and space.This paper proposes a stacked LSTM model combined with the GARCH family,and then combines principal component analysis and entropy model to extract feature factors from multiple influencing factors,and predict the price of apples.The results show that the improved combination model has higher prediction accuracy,which verifies the effectiveness of the improved combination model.