How to remove rows where all the selected columns has NAs with dplyr-CodePudding

I have the following data frame:

dat <- tibble(
  nm = letters[c(1:5)] ,
  g1 = c(1, 1, 2, NA, 2),
  g2 = c(1, NA, 1, NA, 2),
  ga = c(1 :3, NA, 5)
)

It looks like this:

> dat
# A tibble: 5 × 4
  nm       g1    g2    ga
  <chr> <dbl> <dbl> <dbl>
1 a         1     1     1
2 b         1    NA     2
3 c         2     1     3
4 d        NA    NA    NA
5 e         2     2     5

What I want to do is to exclude row when all the column that starts with "g" are NAs. Resulting in this:

 nm       g1    g2    ga
 a         1     1     1
 b         1    NA     2
 c         2     1     3
 e         2     2     5

I tried this but failed:

dat %>% 
  filter_at(vars(contains("g")), !all_vars(is.na(.)))

That gives:

Error in inherits_any():

! Quosures can only be unquoted within a quasiquotation context.

# Bad: list(!!myquosure)

# Good: dplyr::mutate(data, !!myquosure)

CodePudding user response：

What you need is if_all(or if_any), which are supplementary functions to across().

library(dplyr)

dat %>%
 filter( !if_all(starts_with('g'), is.na) )

# # A tibble: 4 × 4
#   nm       g1    g2    ga
#   <chr> <dbl> <dbl> <dbl>
# 1 a         1     1     1
# 2 b         1    NA     2
# 3 c         2     1     3
# 4 e         2     2     5

Equivalently, you could do

dat %>%
 filter( if_any(starts_with('g'), ~ !is.na(.x)) )

dat %>%
 filter( if_any(starts_with('g'), Negate(is.na)) )

CodePudding user response：

You may try

dat %>%
  filter(rowSums(across(starts_with("g"), ~!is.na(.x))) != 0)

  nm       g1    g2    ga
  <chr> <dbl> <dbl> <dbl>
1 a         1     1     1
2 b         1    NA     2
3 c         2     1     3
4 e         2     2     5