基于支持向量机预测的时序多属性决策法  

Decision-making method of time series multiple attribute based on prediction with support vector machine

作　　者：王香柯[1] 王金柱[2] 

机构地区：[1]西安邮电学院应用数学与应用物理系,陕西西安710061 [2]陕西教育学院数理工程系．陕西西安710061

出　　处：《兰州理工大学学报》2010年第2期142-146,共5页Journal of Lanzhou University of Technology

基　　金：陕西省教育厅专项科研计划项目(08JK433)

摘　　要：针对时序多属性决策问题中的预测量大且时间样本有限的特点,以各决策单元的属性时间序列为样本数据,进行多变量时间序列相空间重构,建立用支持向量机预测未来属性值方法,进而提出综合考察决策单元的过去、现在与未来的情况下进行更科学、合理的时序多属性决策方法.Aimed at the feature of decision-making of time sequencing attribute where the prediction amount was large and the time sample was limited in number,a method for prediction of future attribute value was developed based on support vector machine.Multivariable phase space reconstruction was conducted for the time sequencing by taking the attribute time series of several decision-making units as sample data,so that a more scientific and rational method for decision-making of time series multiple attribute was presented,in which the past,present and future condition of decision-making units was comprehensively investigated.

关 键 词：时序多属性决策 多变量时间序列 支持向量机 预测 

分 类 号：O29[理学—应用数学] F201[理学—数学]