河北省奶产量集成预测与实证分析  

Integrated prediction and empirical analysis for milk production in Hebei province

作　　者：刘倩[1] 

机构地区：[1]河北农业大学,河北保定071000

出　　处：《黑龙江畜牧兽医（下半月）》2016年第1期1-3,8,共4页

基　　金：教育部人文社会科学研究规划基金项目(14YJA630075);河北省社会科学基金项目(HB13GL021)

摘　　要：河北省是中国奶产品供给大省之一,近年来河北省奶产量一直处于我国前三名的地位。河北省是否可以持续稳定供给奶产品直接影响到河北省以及京津地区的奶产品整体供给水平,对河北省奶产品进行合理并准确的分析进而进行预测是十分必要的。基于此,笔者提出了一种时间序列及机器学习的集成方法,该方法首先将河北省奶产量时间序列分解为牛奶产量和其他奶产量,在利用ARIMA模型对牛奶产量进行预测建模的同时,使用机器学习方法 -神经网络模型对其他奶产量进行建模,两种模型的集成预测结果即是最终的奶产量预测结果。结果表明:笔者提出的集成模型较传统的预测模型具有更好的预测效果和应用价值。Hebei province is one of the big provinces in the supply of milk products in China,and its milk production has been in the top three positions in China in recent years. Whether Hebei province can sustainably and stably supply milk products directly affects the total supply level of milk products in Hebei province,as well as Beijing and Tianjin areas. It is very necessary to reasonably and accurately analyze and predict the milk products in Hebei province,and the author proposes an integrated method of time series and machine learning based on this. The time- series milk production in Hebei province is firstly decomposed into milk production and other milk production using the method. The other milk production is modeled using neural network for machine learning while using ARIMA model to predict and model milk production. Finally,the integrated prediction results of two models are the final prediction results of milk production. The empirical results show that the integrated model proposed by the author has better prediction effects and application value than traditional prediction models for milk production.

关 键 词：奶产量 时间序列 ARIMA模型 神经网络 集成预测 

分 类 号：S823[农业科学—畜牧学] S851.347.5[农业科学—畜牧兽医]