Методи за въвеждане на липсващи стойности при неигнурируеми механизми

Abstract

In presented research was made an overview of the basic methods for dealing with missing data when nonignorable mechanisms exist. In the beginning different missing data mechanisms are presented. Second part shows several simple methods for Missing data imputation when mechanism is NMAR. In the next sections the basis of the Selection models and the advantages and disadvantages of this models are presented. The final section shows the essence, advantages and disadvantages of the Pattern mixture models.