Kernel-based estimation of P(X > Y) in ranked set sampling

Abstract

This article is directed at the problem of reliability estimation using ranked set sampling. A non-parametric estimator based on kernel density estimation is developed. The estimator is shown to be superior to its analog in simple random sampling. Monte Carlo simulations are employed to assess performance of the proposed estimator. Two real data sets are analysed for illustration. © 2016, Institut d'Estadistica de Catalunya. All rights reserved.