R/purge_values.R

In CorReg: Linear Regression Based on Linear Structure Between Variables

#' Replaces unwanted values by NAs
#' 
#' @description Find values in a dataframe and replace them by NAs. Also give the liste of the variables implied
#' Beware of the factors. The variables stays as factors and the level is still in memory.
#' 
#' @param base the dataframe to clean
#' @param value the value or vector of value to find and remove. if "space" it removes the blank thousands separator.
#' 
#' @return a list with "base" as the cleaned dataset and "list_var" as the list of indexes of variables that have been changed.
#' 
#' @export
purge_values <- function(base, value){
  list_var = c()
  if(value[1] == "space"){#Asci 160
    chr <- function(n) { rawToChar(as.raw(n)) }
    asc <- function(x) { strtoi(charToRaw(x), 16L) }
    value = chr(160)
    for (i in 1:ncol(base)){
      base[, i] = gsub(x = base[, i], pattern = chr(160), replacement = "")
    }
  }
  for(i in 1:ncol(base)){
    vecloc = which(base[, i] %in% value)
    base[vecloc, i] = NA
    if(length(vecloc) > 0){
      list_var = c(list_var, i)
    }
  }
  return(list(base = base, list_var = list_var))
}