Research On Price Forecast Of Shandong Green Onion Based On Time Series Analysis

Price forecasting of agricultural products is an indispensable activity in agricultural economic life,and a research hotspot in the field of agricultural economic management.At present,the related research of agricultural product price prediction mainly focuses on bulk agricultural products,and the research of agricultural product price prediction with regional characteristics is relatively lack.Shandong is the largest agricultural province in China in terms of total output value of agriculture,forestry,animal husbandry and fishery,and has always been known as "China's vegetable basket".Green onion is the most characteristic agricultural product in Shandong Province.Reasonable price prediction can not only maintain the market equilibrium of green onion and other agricultural products,but also promote the stable development of regional agriculture.In the field of time series prediction,research methods are changing with each passing day.The rapid rise of the era of big data has accelerated the development process from traditional agriculture to modern agriculture.With the increasing demand for data analysis in the agricultural field,the use of time series method,intelligent method,multi-scale method and other efficient methods to predict the price of agricultural products has become the focus of today's research.Based on time series analysis,this paper establishes a variety of price forecasting models for green onion in Shandong Province.Considering the effects of a variety of factors to the time series fluctuation tends to be complex and diversified,this paper first constructs the traditional arima-egarch model.For the past few years,as the rise of deep learning and multi-scale methods,the research in the field of time series forecasting is booming.So on this basis,this paper has established LSTM neural network and lstm-ar combination forecasting model based on EEMD to explore their feasibility in the domain of agricultural product price expectation.In the experiment,the validity of the model is tested,and MAE,MAPE and RMSE evaluate the prediction ability.After comparison,the prediction effect of lstm-ar model based on EEMD is better than other models,which can effectively predict the future price of Shandong green onion,and has better prediction precision;however,the traditional ARIMA-EGARCH had higher prediction precision compared with LSTM neural network.