Search code examples
rdplyr

Using pick with rowsums in R

I would like to count the instances where the variables a have been answered (not missing). I tried using pick and rowsums but am unable to find the desired result.

library(dplyr)


anymatch <- function(vars){
  rowSums(pick(all_of(is.na(vars))), na.rm = TRUE)
}


data <- tibble(
  a1 = c("Yes", NA, "No", "Dont know"),
  a2 = c("No", NA, "No", NA),
  a3 = c("Good", "Bad", "Good", NA),
  b1 = c(20, 34, 23, 45)
)

avars <- data |> select(starts_with("a")) |> names()
  

data |> 
  mutate(acount  = anymatch(vars))
Solution

  • a data.table approach

    library(data.table)
setDT(data)[rowSums(is.na(data[, .SD, .SDcols = patterns("^a")])) == 0, ]

    to create a subset. counting/aggregating is easy from here.

    The regex ^a is used to select the needed columns (name starts with a).