CPIT goodness-of-fit tests for the power-law process

被引：7

作者：

Gaudoin, O ^{[1
]}

机构：

[1] LMC, Lab IMAG, F-38041 Grenoble 9, France

来源：

COMMUNICATIONS IN STATISTICS-THEORY AND METHODS | 1998年 / 27卷 / 01期

关键词：

power-law process; non homogeneous Poisson process; goodness-of-fit test; conditional probability integral transformation; reliability growth;

D O I：

10.1080/03610929808832658

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 Power-Law Process is the non homogeneous Poisson process for which the intensity is a power of time. In this paper, we propose a procedure to test the goodness-of-fit of the n first occurrence times of a random point process to the Power-Law Process, based on the Conditional Probability Integral Transformation of O'Reilly-Quesenberry. The new tests are compared through simulation to other tests proposed by Rigdon and Klefsjo-Kumar. It is known that these tests can have a very low power on some alternatives. The major interest of the CPIT tests is that their power is always greater than the previous for these alternatives. Moreover, they ase equivalent to the best tests for large sample sizes.

引用

页码：165 / 180

页数：16

共 25 条

[1]

ASCHER H, 1981, ANN REL MAINT S IEEE, P426

[2] BAYESIAN OPTIMAL OVERHAUL INTERVAL MODEL FOR WEIBULL RESTORATION PROCESS CASE [J].