Incomplete Data Under Different Statistical Estimation And Test Of The Model

In the actual research, missing data has become a universal phenomenon. The existence of missing data often not only makes the error and the variance of the estimator too large and ordinary statistical methods less efficient, but also influences the quality of the statistical data. Therefore, how to deal with the missing data problem has become a hot topic in this field. There are many ap-proaches to study the missing data, which can be classified into two categories: general adjustment method and statistical method specific to a particular statis-tical model. This paper mainly discusses the latter type, we study the parameter estimating methods based on some specific statistical models with missing data. Firstly, the article presents some basic knowledge about missing data, and sim-ply introduces three classical methods and the process of the development of the missing data. In the third part, we use the probability density function and Ker-nel estimating method to discuss the empirical Bayes estimator for parameters, and prove the asymptotic optimality on the ground of the linear exponential dis-tribution model under random censorship. In the fourth part, we introduce two0-1populations with missing at random, and discuss the maximum likelihood estimator. Then the strong of consistency and asymptotic normality property of estimator are proved, finally, the test of the estimator and simulation are discussed.