Short-term Prediction Of Egg Price Based On EEMD-LSTM Combined Model

China is the largest producer and consumer of poultry eggs on the world.The annual output of eggs is nearly 30 million tons,accounting for 43%of the world’s total output.It ranks first in the world for 30 consecutive years.As the second largest industry in China’s livestock products industry,the egg industry plays an important role in the national economy.In recent years,the egg market price fluctuation in China has gradually attracted people’s attention.The long-term and short-term price forecast of traditional agricultural products basically analyzes the fluctuation law of agricultural product prices from the production,consumption and cost of agriculture,and also conforms to the price transmission mechanism of the industrial chain.However,due to the differences in production,processing and eating habits between China,It is dificult to form a complete industrial chain analysis for the study of egg prices.With the development of big data and artificial intelligence technology,more and more researchers have begun to use related technologies such as machine learning to solve the problem of agricultural product price forecasting.In this paper,the short-term prediction of egg prices is mainly studied as follows:1.Get the egg price database.First of all,from the agricultural product price website such as "National Agricultural Products Business Information Public Service Platform",more than 560,000 egg market price data were crawled.By cleaning,sequence normalization,interpolation processing and normalization,price database with time series integrity was constructed based on the egg price data of the agricultural and sideline products wholesale market in Lingjiatang,Jiangsu Province.2.Establish a neural network model for LSTM egg price prediction.Through the Keras framework design and construction of the corresponding LSTM network model,the constructed egg price data set is modeled,and the egg price is predicted through 100 iteration training to obtain the predicted price.Experiments show that the root mean square error(RMSE)and mean absolute percentage error(MAPE)of the LSTM model are 0.24 and 1.68%,respectively,which are 24.91%and 30%lower than the BP model RMSE and MAPE,respectively,compared to the SVR model.They were reduced by 25%and 32.8%respectively.It shows that LSTM neural network is more accurate than BP neural network and support vector machine regression(S VR)model for egg price prediction.3.A prediction method based on EEMD-LSTM combined model is proposed.The set empirical mode decomposition method is studied,and this method is applied to the egg price time series data to obtain 12 IMF components.The corresponding LSTM neural network is constructed for each component,and the obtained prediction results are added to obtain the final predicted value of the combined model.Experiments show that the RMSE and MAPE values of the EEMD-LSTM combination model are 0.16 and 1.44%,respectively,which are 33.33%and 14.29%lower than the RMSE and MAPE values of the LSTM model,respectively.It indicates that the EEMD-LSTM combination model proposed by the paper is more accurate than the single LSTM neural network for the prediction of egg prices.