How do I use the “.” operator in Tidier.jl?

[gioneves]

We can do this in R:

library(dplyr)

df <- data.frame(
a = c(5, 2, 6),
b = c(2, 4, 6),
c = c(10, 2, 1)
)

df %>%
mutate(across((a:c), ~ if_else(. < 5, "OK", "NOK"), .names = "new{col}"))

In Julia, I tried that:

using Tidier

df = DataFrame(a = [5, 2, 6], b = [2, 4, 6], c = [10, 2, 1])

@chain df begin
@Mutate(across((a, b), if_else(.<5, "OK", "NOK") ))
end

But:

ERROR: ParseError:

Error @ REPL[102]:2:31

@chain df begin
@Mutate(across((a:c), if_else(.<5, "OK", "NOK") ))

└┘ ── not a unary operator

Stacktrace:
[1] top-level scope
@ none:1

You can use the “.” operator in Tidier and apply .names (as done in R) to avoid writing them out one by one:

@chain df begin
@Mutate(
newa = ifelse.(a .< 5, "OK", "NOK"),
newb = ifelse.(b .< 5, "OK", "NOK"),
newc = ifelse.(c .< 5, "OK", "NOK")
)
end

[kdpsingh]

Here's what you are looking for:

julia> @chain df begin
         @mutate(across((a, b), x -> if_else.(x .< 5, "OK", "NOK")))
       end

3×5 DataFrame
 Row │ a      b      c      a_function  b_function 
     │ Int64  Int64  Int64  String      String     
─────┼─────────────────────────────────────────────
   1 │     5      2     10  NOK         OK
   2 │     2      4      2  OK          OK
   3 │     6      6      1  NOK         NOK

across() is able to accept anonymous functions like the one shown here. The one caveat with across() is that it operates on entire columns and doesn't auto-vectorize the contents of functions. So when you define the anonymous function, you have to vectorize if_else() as if_else.() and x < 5 as x .< 5.

You can also use a named function, like this:

julia> @chain df begin
         @mutate(across((a, b), function okay(x) if_else.(x .< 5, "OK", "NOK") end))
       end

3×5 DataFrame
 Row │ a      b      c      a_okay  b_okay 
     │ Int64  Int64  Int64  String  String 
─────┼─────────────────────────────────────
   1 │     5      2     10  NOK     OK
   2 │     2      4      2  OK      OK
   3 │     6      6      1  NOK     NOK

Hope that helps, and happy to clarify further.