topic Re: How do I match/merge two datasets using a unique combination of two variables? in SAS Procedures

How do I match/merge two datasets using a unique combination of two variables?

Kels123 — Thu, 21 Jul 2016 19:13:19 GMT

Hi,

I am attempting to merge two databases. To protect confidentiality, one of the databases had all of it's unique identifiers removed, so there is no single variable that is unique for each subject AND present in both databases that I can use for the merge. HOWEVER, I have found that the combination of two date variables (let's call them VAR1 and VAR2) IS unique for each subject.

Is there any easy way to merge based on these two variables (instead of just using one unique variable which I don't have)? My code is below, but I don't think you can list the two variables this way. I sorted on the two variables first.

Proc sort data=clean.INPROGRESS;

by VAR1 VAR2;

run;

proc sort data=clean.ORIGINAL;

by VAR1 VAR2;

run;

data clean.COMBINED;

merge clean.INPROGRESS (in=a) clean.ORIGINAL ;

by VAR1 VAR2;

if a;

run;

Please help! Thank you.

Re: How do I match/merge two datasets using a unique combination of two variables?

LinusH — Thu, 21 Jul 2016 20:13:47 GMT

Haven't you tried it prior to posting?
Doesn't sound correct data wise though to only merge on date variables. Don't you still need some kind of identification of the individuals?

Re: How do I match/merge two datasets using a unique combination of two variables?

ballardw — Thu, 21 Jul 2016 20:15:16 GMT

Run it. It should work assuming your analysis of the unique nature of the paired variables is correct.

BY processing can use a considerable number of variables. The number doesn't change the basic behavior.

Re: How do I match/merge two datasets using a unique combination of two variables?

Kels123 — Thu, 21 Jul 2016 21:25:13 GMT

Hi,

It does work. I just needed to exclude missing VAR1 values from the original databases prior to the merge. Thanks!