Imputation algorithm using copulas

In this paper the author demonstrates how the copulas approach can be used to find algorithms for imputing dropouts in repeated measurements studies. One problem with repeated measurements is the knowledge that the data is described by joint distribution. Copulas are used to create the joint distribution with given marginal distributions. Knowing the joint distribution we can find the conditional distribution of the measurement at a specific time point, conditioned by past measurements, and this will be essential for imputing missing values. Using Gaussian copulas, two simple methods for imputation are presented. Compound symmetry and the case of autoregressive dependencies are discussed. Effectiveness of the proposed approach is tested via series of simulations and results showing that the imputation algorithms based on copulas are appropriate for modelling dropouts.