On the Bayesian Estimation of Gumbel-logistic Unit Distribution under Type-II Censoring

Fatma Wajiha *

Department of Statistics, Patna University, Patna-800005, India.

Manoj Kumar Rastogi

Department of Statistics, Patna University, Patna-800005, India.

Romana Shehla

Department of Statistics, Patna University, Patna-800005, India.

*Author to whom correspondence should be addressed.


Abstract

Unit distributions provide statistical models for observations restricted to the interval (0, 1). This study examines estimation of the parameters and reliability characteristics of the Gumbel–Logistic Unit (GLU) distribution under Type-II censoring. Classical inference is developed through maximum likelihood estimation, with asymptotic confidence intervals derived from the observed Fisher information matrix. Bayesian estimation under the squared error loss function is investigated using Lindley’s approximation and the Metropolis–Hastings algorithm. The latter is also used to construct highest posterior density credible intervals under informative and non-informative prior specifications. A Monte Carlo simulation compares parameter, reliability and hazard estimators for sample sizes of 30, 50, 70, 90 and 110 and observed failure proportions of 50%, 70% and 100%. Performance is evaluated using mean squared errors, interval lengths and coverage probabilities. In the reported scenarios, estimation generally improves with increasing sample size and an increasing proportion of observed failures. Informative-prior procedures frequently yield smaller mean squared errors, while the relative performance of Lindley’s approximation and Metropolis–Hastings estimation depends on the quantity estimated. The approach is illustrated using 20 flood-level observations from the Susquehanna River. The fitted GLU model yields a Kolmogorov–Smirnov statistic of 0.2433 and a p-value of 0.187, supporting a reasonable description of these observations at the 5% significance level.

Keywords: Gumbel-logistic unit distribution, maximum likelihood estimates, lindley approximation, metropolis-hastings algorithm, reliability function, type-II censoring


How to Cite

Wajiha, Fatma, Manoj Kumar Rastogi, and Romana Shehla. 2026. “On the Bayesian Estimation of Gumbel-Logistic Unit Distribution under Type-II Censoring”. Asian Journal of Probability and Statistics 28 (12):36-52. https://doi.org/10.9734/ajpas/2026/v28i12985.

Downloads

Download data is not yet available.