Semi-Parametric and Semi-Nonparametric Estimates of the Confidence Intervals of Quantiles of Physical Quantity Distributions

Abstract

In the present work, new robust adaptive estimates (AE) and confidence intervals have been synthesized for quantiles of physical quantity distributions based on the maximum likelihood method. The AE relative efficiency and a number of the classical and robust estimates have been found by the statistical simulation method on classes of the Tukey local and global supermodels. It is demonstrated that the efficiencies of the adaptive estimates and of the confidence intervals of the distribution quantile are significantly higher than those of the classical parametric, nonparametric, and robust estimates.