topic sort_order in SAS Programming

sort_order

HeatherNewton — Sun, 06 Mar 2022 11:46:22 GMT

data range;
*set pf_lemon;
length score_lbl $9. sort_order 8.;
score_lnl="500<=score<540"; sort_order=1; output;
score_lnl="540<=score<560"; sort_order=2; output;
score_lnl="560<=score<580"; sort_order=3; output;
score_lnl="580<=score<600"; sort_order=5; output;
run;
proc sort data=range; by score_lnl;run;

I tried this but is it how it suppose to be used? do I add a set statement in to include data set I am looking at? I find this very strange.

Re: sort_order

PaigeMiller — Sun, 06 Mar 2022 11:59:44 GMT

Please, from now on, be specific about the desired output. Do not show us the wrong output and then not state what the correct/desired output is (which is what you did). In this case, I can guess. In many other cases, having us guess is not a good way to get the answer you need.

If you want the results sorted by the variable sort_order, then you need to tell PROC SORT to sort by that variable.

proc sort data=range; by sort_order; run;

Re: sort_order

mkeintz — Sun, 06 Mar 2022 16:54:12 GMT

Your program does not produce the results you display. I ran your program:

data range;
  *set pf_lemon;
  length score_lbl $9. sort_order 8.;
  score_lnl="500<=score<540"; sort_order=1; output;
  score_lnl="540<=score<560"; sort_order=2; output;
  score_lnl="560<=score<580"; sort_order=3; output;
  score_lnl="580<=score<600"; sort_order=5; output;
run;
proc sort data=range; by score_lnl;run;
proc print;run;

The proc print produces this:

The SAS System                                                     23:12 Friday, March 4, 2022   8

       score_    sort_
Obs     lbl      order      score_lnl

 1                 1      500<=score<540
 2                 2      540<=score<560
 3                 3      560<=score<580
 4                 5      580<=score<600

which is not the order you display.

Now, I don't recommend using a variable like score_1n1 for sorting purposes, because, as a character variable, it will use lexicographic order. That is, a value like.
'80<=score<=95'

WILL FOLLOW values like
'500<=score<540'.

Better to use the sort_order variable you created.

Re: sort_order

yabwon — Sun, 06 Mar 2022 20:01:39 GMT

Hi Mark,

I generally agree with you recommendation to not to use "score_1n1" like variables for sorting, but just to remind, there is "numeric_collation=on" option available.

data test;
  score_lnl='500<=score<540'; output;
  score_lnl='80<=score<=95'; output;
  score_lnl='1500<=score<1540'; output;
run;

proc SORT
  data=test
  out =test2
  SORTSEQ=LINGUISTIC(NUMERIC_COLLATION=ON);

  by score_lnl;
run;

proc print;
run;

Bart

Re: sort_order

PaigeMiller — Sun, 06 Mar 2022 20:21:14 GMT

Great tip, @yabwon. I keep forgetting about that option myself.

Re: sort_order

mkeintz — Sun, 06 Mar 2022 21:08:04 GMT

... As do I (obviously). Thanks @yabwon