摘要
利用线性回归模型的广义压缩最小二乘估计，引入了有限总体的广义压缩型预测，在预测均方误差意义下，得到了广义压缩型预测优于最佳线性无偏预测的一个充分必要条件；在只能得到每个个体指标的线性组合时，引入了一种线性约束型预测，并得到了线性约束型预测优于最佳线性无偏预测的一个充分必要条件．
The generalized shrunken prediction of finite population is introduced, using generalized shrunken least squares estimator of linear regression models. With respect to prediction mean squared error, a necessary and sufficient condition for superiority of a generalized shrunken prediction over the best linear unbiased prediction is obtained. In the case of linear combination of every unit index, a linear restricting prediction is introduced and then a necessary and sufficient condition for superiority of linear restricting prediction over the best linear unbiased prediction is devived.
出处
《郑州轻工业学院学报》
1999年第1期73-76,共4页
Journal of Zhengzhou Institute of Light Industry（Natural Science）
关键词
线性预报
均方根误差
回归
预测
linear prediction
mean square root error
regression
