Pivoting DataFrame-CodePudding

I have the following dataframe

using DataFrame
df = DataFrame([:a=>["1","1","2"], :b=>["a","b","a"], :c=>[123,321,100]])

Is it possible to pivot it such that each row/column is a distinct value? I.e., pivot it to the following

    "a" "b"
"1" 123 321
"2" 100 missing

CodePudding user response：

This example is almost literally given in help text for unstack function. The only caveat is the confusion caused by giving the column names a, b, c in the example which overlaps the values in df[:b]. Renaming the columns to col_a, col_b, col_c to avoid confusion we have:

julia> df = DataFrame([:col_a=>["1","1","2"], :col_b=>["a","b","a"], :col_c=>[123,321,100]])
3×3 DataFrame
 Row │ col_a   col_b   col_c 
     │ String  String  Int64 
─────┼───────────────────────
   1 │ 1       a         123
   2 │ 1       b         321
   3 │ 2       a         100

julia> unstack(df, :col_b, :col_c, fill=missing)
2×3 DataFrame
 Row │ col_a   a       b       
     │ String  Int64?  Int64?  
─────┼─────────────────────────
   1 │ 1          123      321
   2 │ 2          100  missing

stack and unstack are magical functions and after a while one gets the sense of when they should be used to solve seamingly impossible table jugling tasks.