topic Re: proc compare - show ID in New SAS User

proc compare - show ID

bkq32 — Thu, 09 Apr 2020 21:22:24 GMT

data bweight_old ( keep = patient_id weight--momage );
 format patient_id;
 set sashelp.bweight;
 if _N_ <= 6;
 if _N_ in ( 1, 2 ) then patient_id = "A1";
 if _N_ in ( 3, 4 ) then patient_id = "A2";
 if _N_ in ( 5, 6 ) then patient_id = "A3";
run;

data bweight_new;
 set bweight_old;
 if _N_ = 2 then weight = .;
 if _N_ = 3 then momage = .;
run;

proc compare base = bweight_old compare = bweight_new; 
 ods select CompareDifferences;
run;

In the proc compare output, is there a way for me to see which PATIENT_ID an observation corresponds to?

Re: proc compare - show ID

SASKiwi — Thu, 09 Apr 2020 21:55:34 GMT

proc compare base = bweight_old compare = bweight_new; 
 ods select CompareDifferences;
 id patient_id;
run;

Re: proc compare - show ID

bkq32 — Thu, 09 Apr 2020 22:01:28 GMT

Thank you! I was worried I was doing something wrong because of the warnings in the log about the duplicate IDs, but I guess I'll just ignore them since that's the way the data are structured.

Re: proc compare - show ID

ballardw — Thu, 09 Apr 2020 23:11:37 GMT

@bkq32 wrote:

Thank you! I was worried I was doing something wrong because of the warnings in the log about the duplicate IDs, but I guess I'll just ignore them since that's the way the data are structured.

Unless you are trying to detect an order difference between to data sets you may want to consider sorting both data sets by several variables and have all of those on the ID statement. Multiple variables on the ID require presorting.

Re: proc compare - show ID

bkq32 — Thu, 09 Apr 2020 23:19:37 GMT

Thanks, @ballardw. If I'm understanding this correctly, I think I should be okay since I only have one ID and I'm just taking data set A and running a macro that outputs data set B with the outliers set to null, so the order of the rows should still be exactly the same between A and B.