基于Prophet的生姜价格预测研究  

Research on ginger price forecast based on prophet

作　　者：滕金玲 柳平增[1] 张艳[1] 许世卫 徐光丽[3] 韩蔚[4] Teng Jinling;Liu Pingzeng;Zhang Yan;Xu Shiwei;Xu Guangli;Han Wei(School of Information Science and Engineering,Shandong Agricultural University,Tai'an,271000,China;Key Laboratory of Agricultural Information Service Technology,Ministry of Agriculture and Rural Areas,Beijing,100081,China;School of Mathematics and Statistics,Taishan University,Taian,271000,China;Laiwu Vocational and Technical College,Laiwu,271100,China)

机构地区：[1]山东农业大学信息科学与工程学院,山东泰安271000 [2]农业农村部农业信息服务技术重点实验室,北京市100081 [3]泰山学院数学与统计学院,山东泰安271000 [4]莱芜职业技术学院,山东莱芜271100

出　　处：《中国农机化学报》2020年第8期211-216,共6页Journal of Chinese Agricultural Mechanization

基　　金：农业农村部单品大数据建设项目(11190068);农业部农业信息服务技术重点实验室(SD201801)。

摘　　要：针对农产品价格波动的非线性特征明显、传统时间序列预测方法预测精度不高等状况,构建基于Prophet的农产品价格预测模型,并以生姜为例展开研究。选取2012—2018年生姜每周平均价格数据为研究对象,在对生姜价格的趋势周期分解基础上,通过对生姜价格序列分解的趋势项、周期项和随机项分别进行建模组合实现对2019年上半年生姜价格的预测,并利用统计分析方法对模型性能进行评估。试验结果表明,Prophet算法预测结果的平均相对误差为4%。将Prophet模型的预测结果和BP神经网络预测结果进行比较,其均方误差(MSE为0.20)小于BP神经网络预测结果的均方误差(MSE为0.37)。Prophet预测模型具有较高的预测精度,在农产品价格预测方面具有较广阔的应用前景。In view of the obvious nonlinear characteristics of agricultural product price fluctuation and the low prediction accuracy of traditional time series forecasting methods,the prediction model of agricultural product price based on Prophet was constructed,and ginger was taken as an example to study.Based on the trend periodic decomposition of ginger price,the prediction of ginger price in the first half of 2019 was realized by modeling and combining the trend term,periodic term and random term of ginger price sequence decomposition,and the performance of the model was evaluated by using statistical analysis method.The results showed that the average relative error of prediction results of Prophet algorithm was 4%,and its mean square error(MSE was 0.20)was less than that of BP neural network(MSE was 0.37).The prediction model of prophet has high prediction accuracy and has a broad application prospect in agricultural product price prediction.

关 键 词：价格预测 Prophet算法 神经网络 预测模型 生姜 

分 类 号：F323.7[经济管理—产业经济]