A Comparison of the Prediction of Future Order Statistics for the 2-Parameter Gamma Distribution

This paper considers the prediction of future order statistics based on type-II censored samples from the 2-parameter gamma distribution with unknown scale parameter and semi-unknown shape parameter. Among the four predictors considered, the cross-validatory predictor performed best in terms of guaranteed efficiency. An advantage of this estimator is that it uses only the mean values of order statistics and not the covariances which are more difficult to compute.