Sample data below, say, here is 100 rows/group, need to get the average of X in each group with excluding
the top 10 values?!
data temp; do i=1 to 1000; grp=floor((i-1)/100)+1; x=ranuni(i); output; end; run;quit;
Don't use SQL for this, use PROC RANK (Maxim 14: Use the right tool). The benefit is that PROC RANK makes this simple, and furthermore gives you several options for handling ties.
proc rank data=temp groups=10 out=deciles;
by grp;
var x;
ranks decile;
run;
proc means data=deciles;
by grp;
where decile < 9;
var x;
run;
Don't use SQL for this, use PROC RANK (Maxim 14: Use the right tool). The benefit is that PROC RANK makes this simple, and furthermore gives you several options for handling ties.
proc rank data=temp groups=10 out=deciles;
by grp;
var x;
ranks decile;
run;
proc means data=deciles;
by grp;
where decile < 9;
var x;
run;
Thanks a huge. Anyway to do a list of variables, and save out the means into a dataset?!
Yes, you can use variable lists in both PROC RANK and PROC MEANS. PROC MEANS also allows you to create output data sets.
It's finally time to hack! Remember to visit the SAS Hacker's Hub regularly for news and updates.
Learn how use the CAT functions in SAS to join values from multiple variables into a single value.
Find more tutorials on the SAS Users YouTube channel.
Ready to level-up your skills? Choose your own adventure.