Feature request: function count the missing value

[ynsec37]

Dear developer,

There are some very useful functions that deal with the missing value in the package.
I wonder there is a chance to develop a function like SAS nmiss or cmiss that can
count the missing value.
I had a draft function below referring to the dplyr::coalesce, but it is better that there is a more useful and robust function from your talent developer.

# Count number of missing value
library(dplyr, warn.conflicts = FALSE)
cmiss <- function(..., .blanks_to_na = TRUE) {
  args <- rlang::list2(...)
  if (length(args) == 0L) {
    rlang::abort("`...` can't be empty.")
  }

  args <- vctrs::vec_recycle_common(!!!args)

  stopifnot(length(.blanks_to_na) == 1, is.logical(.blanks_to_na))

  if (.blanks_to_na) {
    args <- purrr::map_if(args, is.character, ~ dplyr::na_if(.x, ""))
  }

  purrr::pmap_int(purrr::map(args, is.na), sum)
}

a <- c(1, 2, NA)
b <- c(3, NA, 4)
c <- "c"
d <- c("NA", "", NA)

# treat "" as `NA` by default
cmiss(a, b, c, d)
#> [1] 0 2 2

cmiss(a, b, c, d, .blanks_to_na = FALSE)
#> [1] 0 1 2

df <- data.frame(v1 = c("a", NA, "b", NA, NA),
  v2 = c(NA, "c", "d", NA, NA),
  v3 = c(letters[5:8], NA),
  v4 = rep(NA, 5))
df %>%
  mutate(n_miss = cmiss(v1, v2, v3, v4),
    first_non_missing = coalesce(v1, v2, v3, v4))
#>     v1   v2   v3 v4 n_miss first_non_missing
#> 1    a <NA>    e NA      2                 a
#> 2 <NA>    c    f NA      2                 c
#> 3    b    d    g NA      1                 b
#> 4 <NA> <NA>    h NA      3                 h
#> 5 <NA> <NA> <NA> NA      4              <NA>

^{Created on 2024-02-24 with reprex v2.1.0}

Thank you very much!

[DavisVaughan]

I think this can actually be done quite elegantly with across() + rowSums()

library(dplyr, warn.conflicts = FALSE)

blanks_to_na <- function(x) {
  dplyr::na_if(x, "")
}

df <- data.frame(v1 = c("a", NA, "b", NA, NA),
  v2 = c(NA, "c", "d", "", NA),
  v3 = c(letters[5:8], NA),
  v4 = rep(NA, 5)
)

df %>%
  dplyr::mutate(across(c(v1, v2, v3), blanks_to_na)) |>
  dplyr::mutate(n_miss = rowSums(across(c(v1, v2, v3, v4), is.na)))
#>     v1   v2   v3 v4 n_miss
#> 1    a <NA>    e NA      2
#> 2 <NA>    c    f NA      2
#> 3    b    d    g NA      1
#> 4 <NA> <NA>    h NA      3
#> 5 <NA> <NA> <NA> NA      4

This also has the benefit of letting you use tidyselect in the across() call to select the columns.

It's not a bad idea to have a psummissing() style function like cmiss(), but I think it is a little too niche for tidyr! Sounds nice for a separate package though!