Estimation of the Largest Success Probability Related to Several Binomial Populations

摘要

We propose several estimators including weighted hybrid estimators and Bayes estimators for estimating the largest success probability. A sequential scheme of sampling is also proposed for the construction of confidence interval for the largest success probability. The length of the confidence interval can thus be controlled. Bias and mean square error of some proposed estimators are studied. Some Monte Carlo results are also attached.