R/dummies_make.R

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

#' @title dummies_make
#'
#' @description Returns the input data with additional dummy variables for
#' specified variables.
#'
#' @param .data A data.frame or tibble.
#' @param .input_variables A character vector containing variables that should
#' be recoded as dummy variables.
#'
#' @return The a data.frame of the input data with additional dummy variables
#' for all .input_variables.
#' @examples -
#' @export

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

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

  # Get dummy levels per variable
  for (i in seq_along(.input_variables)) {

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

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

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

    # Create actual dummy variables in data
    for (j in seq_along(dummy_levels)) {
      name <- paste0(.input_variables[i], dummy_levels[j])
      .data[ncol(.data) + 1] <- NA
      names(.data)[ncol(.data)] <- paste0(name)
      .data[[name]] <- ifelse(.data[[dummy_name]] == dummy_levels[j], 1, 0)
    }
  }

  # Return data with dummy variables
  .data
}