Date:         Thu, 12 Jul 2007 10:56:35 -0400
Reply-To:     Gerhard Hellriegel <gerhard.hellriegel@T-ONLINE.DE>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Gerhard Hellriegel <gerhard.hellriegel@T-ONLINE.DE>
Subject:      Re: PROC SQL: Drop double variables

The only thing I could imagine is to use the dictionary tables for
detecting that situation and build drop-lists to perform. That you could
use as KEEP=(...) option for the input-datasets.
Do you really want to drop ALL dups? What about your KEY-Variables for the
join?
Gerhard


On Thu, 12 Jul 2007 16:34:51 +0200, Will <schimpanski@GMX.DE> wrote:

>Hi,
>
>I'd like to join/merge two large tables a and b where b contains a
>subset of the variables in a (e.g. tid) and other variables.
>
>Example:
>
>Table a                 Table b
>------------            --------------
>id var1 var2            id var1 var3
>1   3.4 10.2            1   3.4 20.1
>2   2.9 11.3            2   2.9 21.3
>3   3.1 13.7            3   3.1 23.5
>4   4.2 10.3            4   4.2 21.1
>
>Table ab
>-----------------
>id var1 var2 var3
>1   3.4 10.2 20.1
>2   2.9 11.3 21.3
>3   3.1 13.7 23.5
>4   4.2 10.3 21.1
>
>
>How can I merge those tables with PROC SQL so the double
>columns/variables in b (e.g. id, var1) are dropped _without_ referring
>to them in the command in an explicit way? (N.B. As I have large
>datasets with several hundred variable explicitly selecting the
>variables would be quite a lot of work.)
>
>At the moment the double variables are kept in the joined dataset which
>causes problems when doing an "outer union corr" concatenation with
>another third table.
>
>Thanks in advance,
>
>Will