Interestingly, one of my colleagues suggested me something that's essentially what you've put down here. It may be a bit challenging to search using an array directly, but if we concatenate all of the indicator variables into one variable, and then use the find() function, it becomes way less complicated. We just need to take note to exhaust all of the possible combinations of the '1111.1' etc etc. Thank you. ... View more

Thank you all for your quick and informative responses to my question! And thank you @Reeza @Jagadishkatam @Kurt_Bremser for pointing out that the by group variables form a hierarchy.   I ran the following to visualize of how the by variables behave differently under different hierarchy. data visualize1; set sample_srt; by member hospital admission; if first.admission then frst_adm = 1; else frst_adm = 0; if last.admission then last_adm = 1; else last_adm = 0; if first.hospital then frst_pos = 1; else frst_pos = 0; if last.hospital then last_pos = 1; else last_pos = 0; run; data visualize2; set sample_srt; by member admission hospital; if first.admission then frst_adm = 1; else frst_adm = 0; if last.admission then last_adm = 1; else last_adm = 0; if first.hospital then frst_pos = 1; else frst_pos = 0; if last.hospital then last_pos = 1; else last_pos = 0; run;   ... View more

It turns out that running the PROC SORT first in this case saved me about 60% of time, as compared to the Option 1 I had. When I ran Option 1, it took a total of about 3 hours to complete, whereas the lines below finished within 1 hour. I haven't compared it with Option 2 yet, but I suspect it will be comparable.   Because the individual state data sets are all sorted by the same variables, I could just do: proc sort data = master out = master_sorted; by var1 var2; run; data NC SC GA FL; set master_sorted; if state = "NC" then output NC; if state = "SC" then output SC; if state = "GA" then output GA; if state = "FL" then output FL; run; Thanks again for your input!   -AH ... View more

@aaronh wrote: Thank you Kurt. So if I run a proc print, what format or informat will it display?.. It will not show you the names of the formats, but it will use the assigned format to display the values. ... View more

Thanks to everyone for your input!   AH ... View more

PROC SQL; CREATE TABLE WORK.foo AS SELECT f.fruit, a.attrib FROM (SELECT DISTINCT fruit FROM USER.HAVE) f FULL JOIN (SELECT DISTINCT fruit, attrib FROM USER.HAVE WHERE attrib NOT IS MISSING) a ON f.fruit = a.fruit; QUIT; ... View more

Thank you so much PG! I was thinking of using a data step merge, but would require a proc sort dedup as well as a lot of renaming and sorting, while this can be achieved in SQL in one step. ... View more

@novinosrin: Thanks for the demo - good to know! Kind regards Paul D.  ... View more

Ok. Got it. Thanks, Reeza! ... View more

@aaronh wrote: Thanks for the input! I was using it in a data step where statement: ... where place = "&place."; Exactly. So if PLACE already has the quotes built in then change that to: where place = &place; ... View more

It works.   Thank you very much Ksharp! ... View more

Almost there. Just swap the x = ii with x = i: data want(drop= i ii); do i = 1 to 10; do ii = 1 to i; x = i; output; end; end; run; ... View more