The Method Of Imputation For Missing Data And It's Application In Stand Growth Model

Missing data is a common phenomena in science research, due to for examplenon-interview, non-response, disqualification of filling or non-existence of original sampleterra in new forest resource inventory, elimination of illogical data in the collection andanalysis process in census, environment monitoring, experimentation in medicine andforest resources inventory etc., gathering and analyzing large information can result inmissing data. When the data is missing, one of the hotspot and the difficulties ininformation statistics analysis is how to deal with the problem scientifically, in order toutilize the information at hand, to reflect the features of the object concemed, andconsequently use these information to predict and forecast to accomplish the expectedpurpose.The main content is as follows:1. Review common imputation methods in dealing with problem of missing data, suchas, single imputation, multiple imputation etc., and introduce emphatically EM arithmetic,DA arithmetic and MCMC arithmetic, into imputation method in statistics, by modernstatistical computation.2. Introduce some typical models in forestry researches. We consider the case whenthe data is incomplete, and deduce an iterative formula based on EM algorithm andMCMC means regarding linear models.3. By taking eight fixed sample fir terra's real data to establish the model, We obtainsome new result concerning applications of three methods mentioned above in standgrowth model for missing data. We also present a parametrical estimation method in thecase of different losing-rate by use of EM arithmetic,Gibbs sampling method ,andmultiple imputation method, It turns out that our method is the best choice for us to solvethe problem when the data is missing.