Você está aqui: Página Inicial / Seminários / 29/06/2018 - Ranked set sampling based estimation in two extended Lindley distributions - Palestrante: Cesar Augusto Taconeli (UFP)

29/06/2018 - Ranked set sampling based estimation in two extended Lindley distributions - Palestrante: Cesar Augusto Taconeli (UFP)

Quando 29/06/2018
das 14h00 até 15h00
Adicionar evento ao calendário vCal
iCal

Seminário Conjunto UFSCar/ICMC


Data e horário: 29/06/2018 às 14 horas

Local: DEs-UFSCar, Sala 43

Título: Ranked set sampling based estimation in two extended Lindley distributions

Palestrante: Prof. Dr. Cesar Augusto Taconeli, Departamento de Estatística, Universidade Federal do Paraná

Resumo: The problem of analyzing lifetime data arises in several applied fields, such as biology, medicine and engineering. Among the distributions used for modeling this type of data, we have the Lindley distribution and some extensions, widely used in the context of reliability studies. Two distributions that generalizes the original one-parameter Lindley distribution were considered in this work: the Power Lindley (PL) and Weighted Lindley (WL) models. Our objective was to develop and evaluate maximum likelihood estimation for PL and WL distributions using three sampling designs based on ranked sets: ranked set sampling (RSS), extreme ranked set sampling (ERSS) and median ranked set sampling (MRSS). Such sampling designs are efficient alternatives to the usual simple random sampling when measuring the variable of interest is difficult or expensive, but it may be possible ranking sample units according to some alternative, cheap and accessible criterion. Through an extensive simulation study, we could assess the best performance of RSS based estimators relative to their simple random sampling counterpart. The impact of different proportions of right censored data was also evaluated. An application based on lifetime of aluminum specimens complements this study.

registrado em: