R/transform.amelia.R

In Amelia: A Program for Missing Data

#' Transform imputed datasets from Amelia objects
#' 
#' Updates the imputed datasets from an \code{amelia} output
#' with the specified transformations.
#'
#' @param _data an object of class "amelia"; typically output from the
#'        function \code{amelia}.
#' @param ... further arguments of the form \code{tag = value}.
#'
#' @details 
#' The \code{\dots} arugments to \code{transform.amelia} are
#' expressions of the form \code{tag = value}, where \code{tag} is the
#' variable that is being updated or created and \code{value} is an
#' expression that is a function of the variables in the imputed
#' datasets. For instance, if you wanted to create an interaction of two
#' imputed variables, you could have one argument be \code{intervar =
#'   var1 * var2}. This would either update the current variable
#' \code{intervar} in the imputed data or append a new variable called
#' \code{intervar} to the imputed datasets.
#'
#' @return
#' An object of class \code{amelia} with its \code{imputations} and
#' \code{missMatrix} values updated according to the transformations. In
#' addition, each of the calls to \code{transform.amelia} are stored in
#'
#' @seealso \code{\link{transform}}
transform.amelia <- function(`_data`, ...) {
  tcall <- match.call(expand.dots = TRUE)
  if (is.null(`_data`$transform.calls)) {
    `_data`$transform.calls <- list(tcall)
  } else {
    `_data`$transform.calls <- c(`_data`$transform.calls, tcall)
  }


  tcall[[1]] <- as.name("transform")
  names(tcall)[2] <- ""
  m <- length(`_data`$imputation)

  orig.data <- remove.imputations(`_data`)
  tcall[[2]] <- orig.data
  new.miss.matrix <- as.matrix(is.na(eval.parent(tcall)))

  for (i in 1:m) {
    tcall[[2]] <- `_data`$imputations[[i]]
    `_data`$imputations[[i]] <- eval.parent(tcall)
  }
  `_data`$missMatrix <- new.miss.matrix
  return(`_data`)
}