IMPROVEMENT IN BIAS AND MSE OF WEIBULL PARAMETER ESTIMATORS FROM RIGHT-CENSORED LARGE SAMPLES BY USING TWO KINDS OF QUANTITIES

摘要

The purpose of the paper is to search for most desirable estimators of the Weibull parameters, which have possibly the smaller |BIAS| and MSE than the conventional ones. We introduce, just as in our recent paper [1], two kinds of quantities:(ⅰ) errors on both axes, and (ⅱ) certain predicted values of unknown data in right-censored large samples. Using these and performing Monte Carlo simulation, we can propose certain estimators for the shape parameter perhaps to achieve our primal purpose adequately.