R/stat.R

In ggdist: Visualizations of Distributions and Uncertainty

# Helper methods for creating stats
#
# Author: mjskay
###############################################################################

# Summarise a data frame using the given function within the specified
# groups, but keep any columns in the groups that have only one value in them
# (i.e. columns where every value in the column is the same)
summarise_by = function(data, by, fun, ...) {
  ddply_(data, by, function(d) {
    new_d = fun(d, ...)
    missing_names = setdiff(names(d), names(new_d))

    # add back in columns with only one value in them
    if (nrow(new_d) > 0) {
      for (col in missing_names) {
        if (length(unique(d[[col]])) == 1) {
          if (is.list(d[[col]])) {
            # list columns must be wrapped
            new_d[[col]] = list(d[[col]][[1]])
          } else {
            new_d[[col]] = d[[col]][[1]]
          }
        }
      }
    }

    new_d
  })
}