Selection Of Predicting Model On The Price Of Agricultural Products In China

After a continuous rise in sugar prices in 2015,at the early time of 2016;CPI grew fast following the pork price soaring.Soaring pork prices dubbed the flying pigs,the flying pigs captured headlines of all the media because of the continue rise of pork price.In recent years,from "Garlic You Relentless","Bean Plays You" to "Pigeon Plays you",and then to "the flying pig",too cheap of the vegetable hurts farmers in our country,it is also happened frequently that people hurts because of too expensive food.The eighteenth National Congress stressed to ensure security of national food and effective supply of main primary products,and to improve the farmers' income,keep the farmers'income growing sustainably and fast,we not only need to control the policy of agricultural products' prices,and need to analysis the tendency of agricultural product price and internal mechanism.To achieve the effective supply of agricultural products,and can promote the farmers' income,Based on the subject of international trade and economic cooperation research institute ministry of commerce commissioned?Study on model selection of monthly price forecast of international agricultural?,put forward model selection to forecast price of agricultural product forecasting in the condition of a single forecast model and combined forecast model.Specifically,the agricultural market prices soaring phenomena as a starting point,the paper introduces the research of agricultural products price movements and the importance of internal mechanism,then briefly introduces the history of agricultural market and expound the characteristics of Chinese agricultural products market and the current main problems of China's agricultural products market.Connecting with the domestic situation,this paper gives the internal and external factors about the market that affect agricultural prices,and selects rice prices,as the representative of agricultural prices.The second,in this paper,starting from the processing of agricultural products price data,this paper introduces the three methods of data processing and the basic principle of three kinds of single forecasting model.Data processing method includes empirical mode(EMD)method,the wavelet transform method,HP(Hodrick-Prescott)filter.Single forecasting model includes BP neural network,grey system prediction and autoregressive moving average model.Selecting the rice prices monthly data as sample,which were obtained from Zhengzhou Commodity Exchange(CZCE)in April 2009-November 2015.As utilizing rice prices data,three kinds of single forecasting model were used respectively for empirical prediction.After that,by comparing with the original sequence,it comes to a conclusion:when a single forecasting model is selected to empirical prediction,neural network prediction effect is the best of all,the effect of grey system forecasting model is the worst.Then,according to the function of data processing method and the characteristics of the single forecasting model,three kinds of combination forecast model is constructed.Combination forecast model includes the HP filter-grey system forecasting model,wavelet neural network prediction model and the EMD-ARIMA model.As utilizing rice prices data,three kinds of combination forecast model were used respectively for empirical prediction in the same way.As utilizing rice prices data,three kinds of combination forecast model were used respectively for empirical prediction in the same way.It comes to a conclusion:when combination forecast model is selected for empirical prediction,the effect of wavelet neural network prediction model is the best of all,the effect of the EMD-ARIMA model is the second.Finally,according to the prediction model of predictive value and actual value comparison chart,absolute error,root mean square error index,after the comparison of single forecast model and combined forecasting model,it draws the following conclusions:"effect of combination forecasting model were better than that of a single prediction model;when involving grey system prediction models,both the result of the single forecasting model and the result of combination of prediction model are poor,but also further proof of the grey system is only suitable for less data and poor information of short-term prediction problem.In addition,through debugging the program and the realization of the process,it also draws the following conclusions:(1)the number of samples in the neural network algorithm than the input layer node number is much larger,implicit layer node number and the dimension of the sample is quite,it can get better prediction accuracy;(2)according to the characteristics of the data,selecting the appropriate data processing methods,then adopting suitable forecasting method,which can be helpful to improve the prediction effect and reduce the prediction error;(3)As for a same forecast model,if the input data is also the same,the longer prediction time,prediction error will be greater.In summary,this paper introduces from the actual economic phenomenon,combining the development history and current situation of agricultural products market,then the agricultural products price formation mechanism and the price of agricultural products impact factors were studied.To find the suitable agricultural products price forecasting model by a plurality of single forecasting model,putting predict results and the prediction error as the evaluation standard,this paper analysis and verifies the feasibility and rationality of the forecasting model.