基于组合模型的我国马铃薯价格预测分析

PRICE FORECASTING OF POTATO WHOLESALE MARKET IN CHINABASED ON COMBINATION MODEL

关键词:  ARIMA模型  GM（1，1）模型  组合预测模型  马铃薯批发市场价格  预测

Key words:  ARIMA  GM (1, 1)  combination forecasting model  potato wholesale market price  forecasting

摘要:

目的 马铃薯是粮菜饲多用途作物，在贫困地区农民增产增收、推进农业供给侧结构性改革和保障国家粮食安全方面发挥着重要作用。对马铃薯价格进行预测分析，对于保障马铃薯生产者利益、稳定马铃薯生产和促进马铃薯产业健康发展具有重要意义。方法 文章利用2005—2019年共15年的马铃薯批发市场价格数据，运用ARIMA预测模型、GM（1，1）预测模型和组合预测模型对2020年马铃薯批发市场价格进行预测。结果 （1）ARIMA预测模型预测精度较高，平均绝对百分比误差在10%以内，可以利用该模型对马铃薯批发市场价格价格进行预测；（2）建立的GM（1，1）预测模型预测效果较好，模型精度为1级，可运用该模型进行价格预测；（3）建立的组合模型预测效果较好，预测精度和稳定性均优于两种单一模型；（4）利用组合预测模型对2020年马铃薯批发市场价格进行预测，结果表明，在2020年马铃薯批发市场价格从高位逐渐回落。结论 2020年马铃薯批发市场价格的下滑态势会对马铃薯市场产生不利影响，马铃薯生产者应根据预测结果主动调整生产策略，控制生产规模，提前制定替代种植计划，以规避价格下滑带来的经济损失。组合预测模型能更好地发挥各单一模型的优势，使预测误差最小化，并可以提供稳定的预测性能，长期预测效果较好。预测模型应根据实际需要不断调整参数和优化模型结构，以实现精准预测。

Abstract:

Potato is a multi-purpose crop, which could be used as food, vegetable and forage, and it plays an important role in increasing farmers' income, advancing structural reform of agricultural supply side and ensuring national food security.The forecasting of potato price is of great significance to protect the interests of potato producers, stabilize potato production and promote the healthy development of potato industry. Based on the price data of potato wholesale market from 2005 to 2019, the ARIMA forecasting model, GM (1, 1) forecasting model and the combination forecasting model were used to forecast the potato wholesale market price in 2020. The first forecasting model, ARIMA model, which was established by using potato wholesale market price data, had a high forecasting accuracy, and the average absolute percentage error was less than 10%. It could be used to predict the price of potato wholesale market. The second forecasting model, the GM （1，1） prediction model, which was established by using potato wholesale market price data, had the good prediction effect and low forecasting error, and the model forecasting accuracy was level 1, which represented a good prediction accuracy, so it could be used for price forecasting. The third forecasting model, the combination forecasting model, had better forecasting effect, and it was better than two single models, and its forecasting results had the best accuracy and stability. The forecasting of the wholesale market price of potato in 2020 was made by using the combination forecasting model. The results showed that the price of the wholesale market in potato gradually dropped from high. The declining trend of potato wholesale market price in 2020 had a negative impact on potato market. Potato producers should actively adjust production strategy, control production scale and make alternative planting plan in advance, so as to avoid the economic loss caused by the declining price. The combination forecasting model can better display the advantages of each single model, minimize the forecasting error, and provide stable forecasting performance. The forecasting model should adjust the parameters and optimize the model structure according to the actual needs to achieve accurate forecasting.