Hybrid estimation and entropy difference test in generalized logistic model for <i>r</i>-largest order statistics

The maximum likelihood estimation (MLE) of generalized extreme value model for r-largest order statistics (rGEV) often overestimates the high quantiles, especially for heavy right tails.To address this problem, Lmoment-based estimation and its hybrid with the MLE for the rGEV were proposed and reported recently, but only the rGEV was studied in the literature.As an alternative to the rGEV, a previous study suggested the generalized logistic model for r-largest order statistics (rGLO).In this study, we investigated some properties of the rGLO, including the entropy difference test.In addition, we applied L-moment-based estimation and its hybrid to the rGLO.The simulation study revealed, for small or moderate samples, that the hybrid estimator showed smaller mean squared error than the other methods including the MLE and L-moment-based estimator.An application to r-largest wind speed data in Korea is provided.