摘要
本文采用主成分分析的方法,对线性模型主成分估算中的误差与风险作了统计分析,得到主成分估计β的平均平方误差不大于模型Y=Xβ+e在添加约束条件P_2β=0下的任一LS无偏估计的平均平方误差。
In this paper the method of main compoent analysis in [ 1 ] is used to make statistics and analysis of error and risk in main component analysis of linear model,The average square error of main component estimation
β obtained from it is not greater than that of any Unbiased estimate LS shown by model y= Xβ+e under the adding of constrant conditionP2β = 0 .
关键词
线性模型
主成分分析
误差
风险
Main component analysis
Modified model
Prediction error
Projection operator Regular model
