SELECTING THE BEST PERFORMING WEIBULL ESTIMATION METHOD FOR RANDOMLY RIGHT CENSORED DATA
Abstract
A total of nine methods are compared and tables are presented for selecting the best performing estimator for a two-parameter Weibull distribution using randomly right censored data. The results indicate that a best performing method can be selected given a certain sample size, censoring level, and shape parameter range. Such a comparison of estimators is necessary as the best performing method is shown to vary across these values. The estimation methods tested include the Maximum Likelihood Estimator, Kaplan-Meier Estimator, Piecewise Exponential Estimator, Földes, Rejt, and Winter Estimator, Klein, Lee, and Moeschberger Partially Parametric Estimator, Ross Estimator, White Estimator, Bain and Engelhardt Estimator, and the Modified Profile Maximum Likelihood Estimator. Recommendations are provided for applying the studied approach to other types of data and distributions.