rank observations by value frequency

[tim bradshaw]

Could anyone please suggest a way to rank groups by the frequency of values in another variable.
In the mock example below, how could I obtain the variable "rankfreq"?

clear
input str1 to_rankfreq    float rankfreq
"a" 1
"a" 1
"a" 1
"b" 2
"b" 2
"c" 3
end

[Nick Cox]

There are various small questions here, notably what you do want to happen if categories tie on frequency.

This works for your data example:

NB I see this as a grouping problem, not a ranking problem in the sense of Stata's rank functions in egen -- they could be used for this problem, but I think there are more direct solutions.

clear
input str1 to_rankfreq    float rankfreq
"a" 1
"a" 1
"a" 1
"b" 2
"b" 2
"c" 3
end

bysort to : gen negfreq = -_N 
egen wanted = group(negfreq to) 

list, sepby(to)

     +----------------------------------------+
     | to_ran~q   rankfreq   negfreq   wanted |
     |----------------------------------------|
  1. |        a          1        -3        1 |
  2. |        a          1        -3        1 |
  3. |        a          1        -3        1 |
     |----------------------------------------|
  4. |        b          2        -2        2 |
  5. |        b          2        -2        2 |
     |----------------------------------------|
  6. |        c          3        -1        3 |
     +----------------------------------------+