Date: Fri, 11 Jan 2008 12:26:34 -0800
Reply-To: "Schwarz, Barry A" <barry.a.schwarz@BOEING.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: "Schwarz, Barry A" <barry.a.schwarz@BOEING.COM>
Subject: Re: data manipulation
In-Reply-To: <251946.53709.qm@web7807.mail.in.yahoo.com>
Content-Type: text/plain; charset="us-ascii"
Use a normal SET statement to read an observation each iteration of the
DATA step. Within the current iteration, use a DO loop with a SET/KEY=
statement to quickly run the chain (e.g., from 1 to 3 to 5 to 8). When
you reach the end of the chain, OUTPUT the final value along with an
incremented counter (such as _n_ from the normal SET). This will produce
observations of the form
1 8 1
2 4 2
3 8 3
5 8 4
7 9 5
To get rid of the intermediate observations, sort this output by new and
counter and then keep only the observations where first.new is 1. You
could even DROP the counter at this point if you want.
-----Original Message-----
From: riya j [mailto:riyaa_j17@YAHOO.CO.IN]
Sent: Thursday, January 10, 2008 7:43 PM
To: SAS-L@LISTSERV.UGA.EDU
Subject: data manipulation
Hi,
I'm trying this for a few days now, but I seem to be going nowhere. I
was wondering if anyone could help me code the logic .
My input data looks something like this
old new
1 3
2 4
3 5
5 8
7 9
so basically, here old and new are old and new ids or accounts or
whatever. The first row means that the old id 1 is changed to a new id
3. Now the id 3 is now 5, as seen in the third obs. This 5 id becomes 8
as per observation number 4. So in the output dataset, i should get only
one row with the old id as 1 and the new id as 8. So basically I'm
compressing the conversion of ids and getting rid of the intermediate
conversions. The second and fifth observation will remain unchanged. The
output data should look something like this
old new
1 8
2 4
7 9
In the above example I showed only 3 intermediate conversions, there
could be many.
| 
