Date:         Mon, 22 Jul 2002 10:23:37 -0500
Reply-To:     Kevin Myers <KevinMyers@AUSTIN.RR.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Kevin Myers <KevinMyers@AUSTIN.RR.COM>
Subject:      Re: merges
Content-Type: text/plain; charset="iso-8859-1"

If it was me, I would consider one of the following alternatives:

A. run PROC TRANSPOSE on the second data set prior to the merge to get all
of the physicians into a single column, OR
B. use an SQL join instead of a data step merge, i.e.:

proc sql;
   create table mylib.merged as
      select * from mylib.table1, mylib.table2
      where table1.physno in(table2.physno1,table2.physno2,table2.physno3);
   quit;

HTH.

Kevin M.


----- Original Message -----
From: "Stacey" <stacey.a.holland@HEALTHPARTNERS.COM>
Newsgroups: bit.listserv.sas-l
To: <SAS-L@LISTSERV.UGA.EDU>
Sent: Monday, July 22, 2002 10:05 AM
Subject: merges


> Hi,
>
> I have 2 datasets that I need to merge but here's the problem:
>
> One dataset contains a physician number and the other datasets also
> contains physician numbers but there are multiple numbers assigned to
> each physician(only in the second dataset)  Right now the dataset is
> laid out with physno1, physno2, physno3 etc all on one record.  What I
> need to do is if the physician number from dataset1 does not match
> with physno1(physician #1) look to see if I can merge it with phys #2
> or phys #3.
>
> Example
> dataset1:                      dataset #2
> phys #                         phy#1      phys#2     phys#3
>
> 22455                          77820      34677      22455
>
> I would need to check each phys # field to find if I have a match.
>
> Any ideas that might work beyond having to merge with each phy #
> field?
>
> Thanks Very Much.