R/getIV.R

In wojciechoblak/varbinq: Simplified processing categorized variables for binary models.

#' Extract a matrix with IV, WOE, structure of bads and structure of goods for each band.
#'
#' @param yVar A logical vector of dependent variable.
#' @param xVar A factor or logical vector of independent variable.
#' @return A matrix with information value, WOE, structure of bads and structure of goods for each band.
#' @examples
#' data(lendclub)
#' getIV(yVar = lendclub$loan_status, lendclub$home_ownership)
#' #or check IV for variable
#' x <- getIV(yVar = lendclub$loan_status, lendclub$home_ownership)
#' sum(x[,"iv"])
#' @export

getIV <- function (yVar, xVar)
{
  # checking ------------------------------------------------------------------
  if(any(length(c(xVar,yVar)) == 0))
    stop("NULL xVar o yVar")

  if(!is.logical(yVar))
    stop("yVar must be logical")

  if(!(is.factor(xVar) | is.logical(xVar)))
    stop("xVar must be factor or logical")

  if(length(unique(yVar)) != 2)
    stop("yVar is constant")


  # amount for each band ------------------------------------------------------
  tBad <- tapply(yVar, xVar, sum) #?li
  tGood <- tapply(!yVar, xVar, sum) #dobrzy

  #erase when number levels of factor are higher than unique values in xVar
  tGood <- tGood[!is.na(tGood)]
  tBad <- tBad[!is.na(tBad)]

  # structure for each band ---------------------------------------------------
  sBad <- tBad / sum(tBad) #struktura z?ych
  sGood <- tGood / sum(tGood) #struktura dobrych

  # Weight of Evidence --------------------------------------------------------
  woe <- ifelse(!is.finite(sBad) | sBad == 0, 0, log(sGood/sBad))

  # Information Value ---------------------------------------------------------
  iv <- (sGood - sBad) * woe

  #return
  cbind(iv, woe, sGood, sBad)
}