Two-sample nonparametric prediction intervals based on random number of generalized order statistics

Abstract

By applying the cumulative hazard transformation, nonparametric prediction, inner and outer intervals based on generalized order statistics (GOSs) are obtained and their exact coverage probabilities are determined. The predictive intervals are accomplished based on informative sample of fixed, as well as random, number of GOSs from a continuous cumulative distribution function (CDF) F. When the sample size is random variable (RV), it is assumed to be positive integer and independent of both informative and future samples. Simulation study and numerical computations are conducted for illustrative purposes.