加权线性回归模型的BLUE影响分析

来源期刊：中南大学学报(自然科学版)2003年第6期

论文作者：邓小青 王志忠

文章页码：722 - 724

关键词：加权线性回归模型;BLUE;W-K统计量;广义相关系数

Key words：weighed regression model; BLUE; W-K statistics; generalized correlation

摘    要：讨论了一组或m组数据扰动或删除前后BLUE的变化及协方差扰动和数据删除对BLUE的混合影响,分别建立了β^{^}_W(J),^β_W(i),^β_{W, gi(j)}与^β_W的关系;提出了度量数据扰动或删除对BLUE的影响的W-K统计量,并给出了W-K统计量与广义相关系数的关系,得到K_I(J)与K_I(I)的上下界;最后用实例说明了W-K统计量在影响分析时的有效性.

Abstract: The changes of the BLUE before and after data-disturbance or data-deletion are considered, and the relations between ^β_W(J),(or ^β_{W(I )}or^β_{W, gi(j)})and ^β_W are established. Next the influences or mixed influences of cases with covariance disturbance and data-deletion on the BLUE are measured by W-K statistics. And then the relationships between W-K statistics and generalized correlation are found, and the ranges of K_I(J)and K_I(I) are gained. In the end, an example is provided for illustration purpose. The results show that this method is useful in practice.