CONFIDENCE BANDS FOR LINEAR-REGRESSION OVER RESTRICTED REGIONS

被引：0

作者：

SEPPANEN, E ^{[1
]}

UUSIPAIKKA, E ^{[1
]}

机构：

[1] UNIV TEXAS, AUSTIN, TX 78712 USA

来源：

SCANDINAVIAN JOURNAL OF STATISTICS | 1992年 / 19卷 / 01期

关键词：

RESTRICTED PREDICTOR VARIABLES; CONVEX REGIONS; HYPERBOLIC CONFIDENCE BANDS;

D O I：

暂无

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

020208 ; 070103 ; 0714 ;

摘要：

The most common method for constructing a hyperbolic confidence band for linear regression is the Scheffe procedure. However, if we are only interested in restricted regions for the values of the predictor variables, this procedure is conservative and can yield unnecessarily wide bands. In this paper we first consider exact confidence bands when the region of interest is determined in a certain sense by a subspace of the prediction space. Then we use these bands for constructing conservative confidence bands for some convex sets of the prediction space.

引用

页码：73 / 81

页数：9