Distribution free testing for conditional distributions given covariates

被引：4

作者：

Khmaladze, Estate ^{[1
]}

机构：

[1] Victoria Univ Wellington, Sch Math & Stat, POB 600, Wellington, New Zealand

来源：

STATISTICS & PROBABILITY LETTERS | 2017年 / 129卷

关键词：

Unitary operators; Empirical process; Kiefer process; Parametric models; Orthogonal projections;

D O I：

10.1016/j.spl.2017.06.026

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：

Given a sample (xi(i),X-i)(i=1)(n) of i.i.d. pairs, consider F-x and Q(x) as two different models for conditional distribution of ?i given X-i = x. The paper shows how the empirical process for testing F-x can be transformed into a process with the same asymptotic behavior as the empirical process for testing Q(x) thereby rendering the two testing problems equivalent. (C) 2017 Elsevier B.V. All rights reserved.

引用

页码：348 / 354

页数：7

共 17 条

[1] Distribution-free specification tests of conditional models [J].