Developer function to add number of NAs to evaluator results

A helper function to append rows with number of NAs (per group, if applicable) to evaluator results tibble.

Usage

add_na_counts(out, data, value_col, na_rm, ...)

Arguments

out

Evaluator results tibble to append number of NA results to.

data

Data used to compute number of NAs.

value_col

Character string, specifying the column used to compute the number of NAs.

na_rm

A logical value indicating whether NA values should be stripped before the computation proceeds.

...

Additional name-value pairs to pass to dplyr::mutate() to append columns.

Value

Tibble with additional rows containing the new metric "num_na" and its corresponding ".estimate"

Examples

# generate example fit_results data with NA responses
fit_results <- tibble::tibble(
  .rep = rep(1:2, times = 2),
  .dgp_name = c("DGP1", "DGP1", "DGP2", "DGP2"),
  .method_name = c("Method"),
  # true response
  y = lapply(1:4, FUN = function(x) c(rnorm(100 - x), rep(NA, x))),
  # predicted response
  predictions = lapply(1:4, FUN = function(x) rnorm(100))
)

# evaluate root mean squared error and number of NA responses for each row in
# fit_results
rmse_na_fun <- function(data, truth_col, estimate_col, na_rm = TRUE) {
  out <- tibble::tibble(
    .metric = "rmse",
    .estimate = yardstick::rmse_vec(
      data[[truth_col]], data[[estimate_col]], na_rm = na_rm
    )
  ) %>%
    add_na_counts(data = data, value_col = truth_col, na_rm = na_rm)
  return(out)
}
eval_results <- eval_constructor(
  fit_results = fit_results,
  fun = rmse_na_fun,
  truth_col = "y",
  estimate_col = "predictions",
  na_rm = TRUE
) %>%
  tidyr::unnest(.eval_result)