Conditional Inference on the Extreme Value Type-I Distribution Based on Generalized Progressive Hybrid Censored Scheme

Abstract

Abstract It is widely known that conditional inference is usually just as effective as Bayesian inference based on a non-informative prior. However, it is less efficient than Bayesian inference based on the informative prior distribution. Therefore, the main objective is to find the conditional point estimates using the pivotal functions for the extreme value type-I distribution parameters, based on the generalized progressive hybrid-censoring scheme, and compare them with the Bayesian estimates, via Monte Carlo simulation. The simulation results showed that conditional inference is highly efficient and provides better estimates than Bayesian estimates based on different loss functions. Finally, weather phenomenon data has been analyzed to demonstrate the efficiencies of the proposed methods and to ensure the importance of this model for analyzing this data. AMS Classification: 62F10, 62F15, 62H10.