Prediction of record values by using quantile regression curves and distortion functions

Abstract The purpose of the paper is to provide a general method based on conditional quantile curves predict record values from preceding records. predictions are median (or regression) curves. Moreover, quantiles used confidence bands for these predictions. recently introduced concept multivariate distorted distributions that instead copulas represent dependence structure. This allows us compute in simple way. theoretical findings illustrated with non-parametric model (standard uniform), two parametric models (exponential and Pareto), procedure case. A real data set simulated case study reliability analysed.