dlsem (version 1.5)

EM.imputation: Imputation of missing values using Expectation-Maximization

Description

Missing values of quantitative variables are replaced by their expected value computed using the Expectation-Maximization (EM) algorithm (Dempster et al., 1977). A Multivariate Gaussian distribution is assumed.

Usage

EM.imputation(x = NULL, group = NULL, data, plotDir = NULL, tol = 0.0001, maxit = 500, quiet = FALSE)

Arguments

x

A vector containing the names of the quantitative variables to be imputed. If NULL, all quantitative variables will be imputed.

group

The name of a group factor (optional). If NULL, no groups are considered.

data

An object of class data.frame containing the variables to be imputed and, if group is not NULL, the group factor.

plotDir

A directory where to save the plots after imputation (optional). If NULL, no plots will be produced.

tol

The tolerance threshold of the EM algorithm. Default is 0.0001.

maxit

The maximum number of iterations for the EM algorithm. Default is 500.

quiet

A logical value indicating whether messages on the status of the algorithm should be shown. Default is FALSE.

Value

data.frame provided to argument data, where the requested variables have been imputed.

References

A. P. Dempster, N. M. Laird, D. B. Rubin (1977). Maximum Likelihood from Incomplete Data via the EM Algorithm. Jornal of the Royal Statustical Society, Series B, 39(1), 1-38.

See Also

unirootTest, applyDiff

Examples

data(industry)
EM.imputation(c("Consum","Pollution"),group="Region",data=industry)