您的位置:首页  > 论文页面

一种改进的组合预测模型在猪瘟月新发生次数中的应用

发表时间:2013-01-15  浏览量:1949  下载量:803
全部作者: 毛泽强,焦桂梅
作者单位: 兰州大学数学与统计学院
摘 要: 以贵州省2007年至2010年的猪瘟(classical swine fever, CSF)月新发生次数为训练数据,最初分别用灰色系统模型(gray model)[GM(1,1)]和求和自回归滑动平均(auto regressive integrated moving average)[ARIMA(p, d, q)]模型进行拟合,用2011年前11个月的实际猪瘟月新发生次数与预测数据进行比较和模型评价,然后基于这2个模型,提出2种改进模型:赋权组合模型和残差修正模型。通过模型预测效果评价指标,最终选择残差修正模型,其预测效果显著于单独用GM(1,1)或ARIMA(p, d, q)模型以及赋权组合模型。这对预测离散的整值数据问题提供了一种有效的方法。
关 键 词: 时间序列分析;灰色预测;GM(1,1);ARIMA(p,d,g);组合模型
Title: A corrected hybrid approach for predicting the number of monthly new outbreaks of CSF
Author: MAO Zeqiang, JIAO Guimei
Organization: School of Mathematics and Statistics, Lanzhou University
Abstract: In this paper, the data of monthly new outbreaks of classical swine fever (CSF) in Guizhou province from January 2007 to December 2010 were used for model training, GM(1,1) and ARIMA(p, d, q) models were used for the fitting. And the data of autual CSF outbreaks of the first 11 months in 2011 were used to test the effectiveness of the models. Based on the two models, this paper gave two methods (weighted combination model and residual error correction model) to improve the above models. Through the model prediction effect evaluation, residual error correction model was selected since its prediction effect was better than GM(1,1), ARIMA(p, d, q) and weighted combination model respective. This provides an effective method for the prediction of the discrete integer datum.
Key words: time series analysis; gray prediction; GM(1,1); ARIMA(p,d,g); combined model
发表期数: 2013年1月第1期
引用格式: 毛泽强,焦桂梅. 一种改进的组合预测模型在猪瘟月新发生次数中的应用[J]. 中国科技论文在线精品论文,2013,6(1):23-31.
 
0 评论数 0
暂无评论
友情链接