R/dummies_annotate.R

In janbrederecke/prepareD: Functions that help to prepare data before analysis

#' @title dummies_annotate
#'
#' @description Returns the annotation with additional dummy variables for
#' specified variables. Requires a data.frame or matrix object of name
#' 'annotation' in the global environment. The annotation object must be
#' specified using the column names "name", "pname", "unit", "short_pname", and
#' "comment". This function grabs the "pname" of the .input_variables to create
#' the dummy names.
#'
#' @param .data A data.frame or tibble.
#' @param .input_variables A character vector containing variables that should
#' be recoded as dummy variables in the annotation.
#'
#' @return The annotation with additional dummy variables for all
#' .input_variables.
#' @examples -
#' @export

dummies_annotate <- function(.data
                             , .input_variables
){

  # Check if all .input_variables are in data
  if (sum(.input_variables %in% names(.data)) < length(.input_variables)) {
    stop("Some variables are not in data!")
  }

  # Check which variable needs dummies and add to annotation
  for (i in seq_along(.input_variables)) {

    # Find out which number of names(.data) corresponds to the variable
    num <- which(names(.data) %in% .input_variables[i])

    # Check variable type and define dummy_levels accordingly
    if (is.factor(.data[[num]]) == TRUE |
        is.character(.data[[num]]) == TRUE) {
      dummy_levels <- as.character(unique(stats::na.omit(.data[[num]])))
    }

    if (is.numeric(.data[[num]]) == TRUE) {
      dummy_levels <- as.numeric(as.character(sort(unique(
        stats::na.omit(.data[[num]])
      ))))
    }

    # Print the levels of every variable
    print(paste0("Dummy levels of ", names(.data)[num], ": ",
                 paste(dummy_levels, collapse = ", ")))

    # Create the actual new names for the dummy variables
    for (j in seq_along(dummy_levels)) {
      name <- paste0(.input_variables[i], dummy_levels[j])

      # Check variable type and define dummy variable name accordingly
      if (is.factor(.data[[num]]) == TRUE |
          is.character(.data[[num]]) == TRUE) {
        pname <-
          paste0(annotation[[2]][which(annotation[[1]] %in%
                                       names(.data)[[num]])],
                 ":",
                 dummy_levels[j])
      }

      if (is.numeric(.data[[num]]) == TRUE) {
        pname <-
          paste0(annotation[[2]][which(annotation[[1]] %in%
                                       names(.data)[[num]])],
                 ":",
                 dummy_levels[j])
      }

      # Create additional row and add it to annotation
      unit <- short_pname <- comment <- NA
      new_row <- cbind(name, pname, unit, short_pname, comment)
      annotation <- rbind(annotation, new_row)
      rownames(annotation) <- annotation[[1]]
    }
  }

  # Return ammended annotation
  annotation
}