R/05-quintile_ppi.R

In ernestguevarra/ppicalc: Calculator for Determining Poverty Likelihood Based on the Poverty Probability Index (PPI)

################################################################################
#
#' Group population by quintiles based on their poverty probability
#'
#' @param ppi A numeric vector of poverty probabilities or likelihoods.
#' @param probs A numeric vector between 0 to 1 of quantile probabilities.
#'     Default values are a vector of equal probabilities (0.2, 0.4, 0.6, 0.8, 1).
#' @return A numeric vector between 1 to 5 classifying data into quintiles of
#'     increasing poverty probabilities.
#'
#' @examples
#' # ppi based on USAID extreme poverty definition using Bangladesh data from
#' # washdata package and PPI lookup table from ppitables package
#' # install.packages("washdata"); library(washdata)
#' ppi <- get_ppi_cohort(index = washdata::indicatorsDataBGD$ppi,
#'                       ppiTable = ppitables::ppiBGD2013)
#' quintile_ppi(ppi = ppi$extreme)
#'
#' @export
#'
#
################################################################################

quintile_ppi <- function(ppi, probs = c(0.2, 0.4, 0.6, 0.8, 1)) {
  #
  #
  #
  if(class(probs) != "numeric") {
    stop("Quantile probabilities must be a numeric vector. Try again.")
  }

  if(any(probs > 1) | any(probs < 0)) {
    stop("Quantile probabilities must be between 0 and 1. Try again.")
  }
  #
  # Get quintile cutoffs
  #
  qCutOff <- quantile(x = ppi, probs = probs)
  #
  # Classify households by wealth quintile
  #
  pQuintile <- ifelse(ppi <= qCutOff[1], 1,
                 ifelse(ppi > qCutOff[1] & ppi <= qCutOff[2], 2,
                   ifelse(ppi > qCutOff[2] & ppi <= qCutOff[3], 3,
                     ifelse(ppi > qCutOff[3] & ppi <= qCutOff[4], 4, 5))))
  #
  # return output
  #
  return(pQuintile)
}