期刊名称：Brazilian Journal of Probability and Statistics
印刷版ISSN：0103-0752
出版年度：2003
卷号：17
期号：1
页码：1-16
出版社：Brazilian Statistical Association
摘要：Robust Bayesian Inference in a linear-linear segmented re-gression model, assuming non-homogeneous error variance, is explored forcases in which the errors follow Student's t distributions. Metropolis-within-Gibbs algorithms are used in order to estimate the posterior distribution ofthe change point and the model parameters. The methodology is illustratedby the analysis of two simulated data sets and a real data set from a clinicalstudy designed to determine the anaerobic threshold of a healthy male duringdynamic exercise
