Anonymous inline functions

[ymer]

This code works at extracting the number:

df = DataFrame(ID1 = ["ID_1", "ID_23", "ID_456"])
extract_id(s) = split(s, "_")[end]
df = @mutate(df, ID = extract_id.(ID1))

But I would like to be able to do it without the helper function. However, this does not work:

df = @mutate(df, ID = (s -> split(s, "_")[end]).(ID1))

[kdpsingh]

I'll take a look at this! Thank you for posting.

[kdpsingh]

The challenge here is that TidierData thinks that s refers to a column name in the DataFrame. This is one of the downsides of using bare column names to refer to columns. One thing on the roadmap is to implement some dynamic scoping, whereby TidierData will "detect" that there is no column called s and will rewrite the code in a way that will make this example work.

For now, the best way to work with anonymous functions in TidierData is to use across():

julia> df = DataFrame(ID1 = ["ID_1", "ID_23", "ID_456"])
3×1 DataFrame
 Row │ ID1    
     │ String 
─────┼────────
   1 │ ID_1
   2 │ ID_23
   3 │ ID_456

julia> df = @mutate(df, across(ID1, s -> [split.(s, "_")[end] for s in s]))
3×2 DataFrame
 Row │ ID1     ID1_function 
     │ String  SubString…   
─────┼──────────────────────
   1 │ ID_1    1
   2 │ ID_23   23
   3 │ ID_456  456

[kdpsingh]

Here's another approach that works:

julia> df = @chain df begin 
                @mutate(across(ID1, s -> split.(s, "_")))
                @unnest_wider(ID1_function)
            end
3×3 DataFrame
 Row │ ID1     ID1_function1  ID1_function2 
     │ String  SubString…     SubString…    
─────┼──────────────────────────────────────
   1 │ ID_1    ID             1
   2 │ ID_23   ID             23
   3 │ ID_456  ID             456