R/variable_summary.R

In TriDimRegression: Bayesian Statistics for 2D/3D Transformations

#' Computes mean and optional probabilities for a given
#' variable.
#'
#' @param var_name String, name of the dependent variable
#' @param var_matrix Numeric matrix samplesN x observationsN.
#' @param probs A numeric vector of quantiles to compute.
#'
#' @return data.frame with "{var_name}" column for the mean and
#' "{var_name}_{prob}" columns for each probability.
#' @keywords internal
#' @export
#' @examples variable_summary("test", matrix(1:1000, ncol = 1), c(0.05, 0.95))
variable_summary <- function(var_name, var_matrix, probs){
  # sanity checks
  stopifnot(is.character(var_name), !is.na(var_name))
  stopifnot(sum(is.finite(var_matrix)) == length(var_matrix))
  stopifnot(is.numeric(probs) || is.null(probs))

  var_summary <- list()
  var_summary[[glue::glue("{var_name}")]] <- apply(as.matrix(var_matrix), MARGIN=2, FUN=mean)
  if (is.null(probs)) return(as.data.frame(var_summary))

  computed_probs <- purrr::map(probs, ~apply(as.matrix(var_matrix), MARGIN=2, FUN=quantile, probs=.))
  names(computed_probs) <- glue::glue("{var_name}_{probs*100}")
  as.data.frame(c(var_summary, computed_probs))
}