Market Price Analysis And Prediction Of Qianxi Chestnut Based On Neural Network Model

Qianxi chestnut in the international market has the "Oriental Pearl" reputation.Qianxi chestnut is abundant in Qianxi area of Tangshan City,Hebei Province,and it is characteristic agricultural product of Hebei Province.From the perspective of traditional Chinese medicine,chestnut has beneficial functions such as stomach calming and anti-arterial sclerosis,so it is favored by the majority of consumers.In recent years,Qianxi chestnuts sell well in more than 200 large and medium-sized cities and countries and regions.Its exports in Hebei Province accounted for nearly 1/2 of the total export volume.Chestnut industry has become the "green bank" of farmers in Qianxi County.Its stable and healthy development has important relation with the economic development of Hebei Province.Over the years,many scholars at home and abroad have done research on chestnut industry,but most scholars study chestnut cultivation technology,breeding technology,pest control technology and chestnut storage and deep processing technology.Based on economic point of view,using macro microeconomic theory and statistical methods to study chestnut price fluctuations and predictions of literature is very few.So this paper is based on nonlinear autoregressive neural network technology,adding economic theory:economic cycle,supply and demand theory,elasticity theory,cobweb model theory,regional brands and geographical signs theory guidance,creating models for Qianxi chestnut price data and predicting price data dor the next 5 issues.In this paper,the price data of the Qianxi chestnut small grain,Qianxi chestnut mid-grain,the large grain of Qianxi chestnut are constructed by the method of time series model and neural network model.For the time series model:the price sequence of the Qianxi chestnut small grain is stable,so the AR(1)model is established after analysis and testing.The first order difference data of the Qianxi chestnut large grain is stable.After analysis and test,so the ARIMA(1,1,1)model is established.Respectively,the price data of two sequences of Qianxi chestnut small grain and Qianxi chestnut large grain.Because the Qianxi chestnut mid-grain sequence is not smooth,the first order difference is white noise sequence,so it is not suitable to establish a time series model.Building a nonlinear self-regression neural network model,after many trainings,according to the selected model evaluation indicators,choose a better neural network model.Then bring the data to make three price sequence predictions.Introducing the concept of combined model,determine the weight of the time series model and neural networkmodel in the combined model,making the price predict data of the combined model.Then compare the prediction effect of neural network model,time series model and combination model.The neural network model is better than the combination model,and the combination model is better than the time series model.The neural network model is finally selected,and then predict for the next 5 periods price data of the Qianxi chestnut small grain,the Qianxi chestnut mid-grain,and the Qianxi chestnut large grain.According to the forecast,the price of Qianxi Chestnut small grain will fluctuate between 12.6 yuan a kilo and 15.6 yuan a kilo in the next 5 months.The price of Qianxi Chestnut mid-grain will fluctuate between 15.2 yuan a kilo and 18.4 yuan a kilo in the next 5 months.The price of Qianxi Chestnut large grain will fluctuate between 15.0 yuan a kilo and 31.2 yuan a kilo in the next 5 months.