Result Details
On the confidence intervals for mean of left-censored Weibull distribution
Michálek Jaroslav, doc. RNDr., CSc., FME (FME)
Left-censored data with two detection limits occur frequently in many application areas, especially in environmental sciences. This paper suggests a computational procedure for calculation of confidence intervals for the expected value of type I doubly left-censored Weibull samples. Three methods of confidence bounds construction are proposed, specifically the maximum likelihood method, and parametric and non-parametric bootstrap. Performances of the particular methods are assessed using simulations. It is shown that bootstrap-based confidence intervals have less satisfactory coverage properties for small sample sizes and high level of censoring.
Bootstrap, doubly left-censored data, expected value, Fisher information matrix, maximum likelihood.
@inproceedings{BUT125961,
author="Michal {Fusek} and Jaroslav {Michálek}",
title="On the confidence intervals for mean of left-censored Weibull distribution",
booktitle="MENDEL 2016, 22nd International Conference on Soft Computing",
year="2016",
journal="Mendel Journal series",
number="22",
pages="249--254",
address="Brno",
isbn="978-80-214-5365-4",
issn="1803-3814"
}