Study On The Mechanism Of Fluctuation And Comparison Of Prediction Methods For The Pig Price In China

The stability of pig price directly relate to the stable development of national economy,because the majority of meat consumer goods is pork in China.Therefore,it is particularly important to explore the characteristics and forming mechanism of price fluctuation for pig in China.On that basis,building effective system of pig price prediction is very important meaning for the steady development of pig industry.Firstly,we analyzed the overall trend of price,fluctuation period,seasonal fluctuation and random fluctuation of Chinese pig price using time series decomposition method.We studied the main factors affecting the fluctuation of living pig price based on the theory of supply and demand and factor analysis method.Then,we used five methods including multiple linear regression,support vector machine(SVM),ARIMA,neural network model and combination forecasting to forecast pig prices in short-term,medium-term and long-term respectively in a specific period,and the best forecasting period of each forecasting model was determined by empirical analysis.The main results are listed as follow.(1)From January 2000 to July 2018,Chinese national pig price experienced five cycles of fluctuations,each cycle lasting 40 months or more.Cycle fluctuation,seasonal component and random component are the main factors causing the fluctuation of pig price.(2)The piglet price and the breeding profit represented by the breeding profit have a significant impact on the fluctuation of the pig price;the change of pig yield is still the direct cause of the change of pig price.The change of production cost based on feed is the basic reason of the change of pig price.The consumption of animal food tends to be diversified,which leads to the obvious influence of pork substitutes on the price of living pigs.(3)In different degree of fluctuation,combination forecasting has the highest prediction accuracy and prediction accuracy for different forecast period pork prices.Among them,ARIMA model is the best in short-term and medium-term prediction under the condition of intense fluctuation,and BP neural network is the best in longterm prediction.Under the condition of gentle fluctuation,the forecasting effect and accuracy of several forecasting methods are very high in short-term,medium-term and long-term forecasting.In addition,the prediction effect and accuracy of each prediction method are better under the condition of gentle fluctuation than under the condition of severe fluctuation.However,with the increase of forecasting period,the forecasting effect and accuracy of various forecasting models also decrease.