R/compare_distributions.R

In sarahvanhala/nomscheck: Report to Check NOMS Data

#' @import ggplot2
NULL

#' Compare Distributions
#'
#' The purpose of this plot is to search for outliers/errors/typos in client-specific H-data values.
#'
#' Simulates a gamma distribution and compares client-specific H-data values to the simulated distribution.
#'
#' If optional variables are not specified, the function will use a reference value.
#'
#' @param newdata Dataframe downloaded from SPARS containing all assessments that you want to check. This will usually just include intakes.
#' @param var H-data variable of interest
#' @param mean (optional) mean of the distribution of the variable of interest
#' @param sd (optional) standard deviation of the distribution of the variable of interest
#' @param binwidth (optional) The binwidth of the plot
#' @return Histogram of simulated data with lines indicating client's actual H-data values.
#' @examples
#' compare_dist(newdata, var, mean, sd, binwidth)
#' @export
compare_dist <- function(newdata, var, mean, sd, binwidth) {
  if (!(var %in% names(refdist))) {
    stop("var must be one of the following: ", paste(names(refdist), collapse = ", "))
  }
  if (missing(mean)) {
    mean <- refdist[[var]][["mean"]]
  }
  if (missing(sd)) {
    sd <- refdist[[var]][["sd"]]
  }
  if (missing(binwidth)) {
    binwidth <- refdist[[var]][["binwidth"]]
  }
  alpha <- mean ^ 2 / sd ^ 2
  beta <- sd ^ 2 / mean
  simdata <- tibble(x = rgamma(1000, shape = alpha, scale = beta))
  ggplot() +
    theme_minimal() +
    geom_histogram(data = simdata, aes(x = x), color = "#d9d9d9", fill = "#d9d9d9", binwidth = binwidth) +
    geom_vline(xintercept = newdata[[paste0("der_", stringr::str_to_lower(var))]], color = "#083D55") +
    theme(axis.title = element_blank(), axis.ticks.y = element_blank(), axis.text.y = element_blank()) +
    ggtitle(paste("Distribution of new values of", var),
            subtitle = "Compared to reference distribution")
}