Quantile Estimation Of Missing Data Based On Gamma Distribution

Missing data widespread in the experimental research and social investigation,data missing results in the aggravation of the analysis task and the inaccuracy of the s-tatistical results.How to effectively deal with the missing data, how to make full use of the data and accurately reflect the characteristics of the study colony, to achieve the desired purpose, has become a difficult and hot problem in current statistical research. Due to the missing data often contains some important information, wish to the statistical analysis method applied to deal with the missing data, the general idea is to fill the incomplete data, and get the complete data set, and then the data set was analyzed, the conclusions in order to ensure the accurate response to the real situation.In this paper, we study the problem of missing data in the not missing at ran-dom(NMAR) mechanism, under some assumptions, by the observed data, using the method of probability statistics give out supplementary data, the statistical estima-tion of the distribution parameters and quantile estimation of probability distribution of complete data is given by the moment estimation. We are more concerned with the median, quantile. At the same time, we obey the Gamma distribution and the probability of success is exponential, the parameters and quantiles of the Gamma distribution are estimated,and the iterative formula of the parameters is given. In the simulation period rely on R language the missing ratio of 1/10, 1/5, 1/2, 2/3 were simulated respectively. The simulation results are compared with each other,and good results are obtained. Most of the time the two parameter model can not meet the actual needs so we extend the model to the three parameter, and the itera-tive formula of the parameters is given and the problem of choosing the parameters of the probability of success is explained. To deal with some of the missing data to give some help.