The Statistical Analysis Of Incomplete Data

In this paper, we talk mainly about statistical analysis of censored samples. Censored samples may arise in practice in different way. for example, in life testing experiments. We talk about it in three parts.l.Observables, a general class of distributions, are to be predicted using the Bayesian approach. The class of distributions includes the weibull, burr-type xⅡ, pareto and beta distributions. A proper general prior density function is suggested, and predictive density functions are obtained in one-and two-sample cases when the history sample is a type II double censored sample. Illustrative examples are given.2. We also obtain the Bayes estimators of the Burr type II distribution parameter, and we give a simple approximation.3.we obtain a kind of approximation of the likelihood function, and we show that the approximation is equivalent to the exact one when the sample size is large. Based on this approximation, we obtain the approximate maximum likelihood estimation of parameters of logistic distribution.4.In this part, we obtain the empirical Bayes estimation for the parameter of the gamma distribution families Y(θ,1/2), and obtain convergent rate.