Date:         Sun, 29 Jun 2008 19:27:54 -0400
Reply-To:     Muthia Kachirayan <muthia.kachirayan@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Muthia Kachirayan <muthia.kachirayan@GMAIL.COM>
Subject:      Re: sort
Comments: To: "Raynaud, Scott" <Scott.Raynaud@targetbase.com>
In-Reply-To:  <D6EA6D1B6E318A48B30521DE8C6CC6960864A82D@DMSP-MSG-EVS01.mail.pvt>
Content-Type: text/plain; charset=ISO-8859-1

On Sun, Jun 29, 2008 at 5:48 PM, Raynaud, Scott <
Scott.Raynaud@targetbase.com> wrote:

> Proc summary has been suggested as an alternative.  My data look like
> this:
>
> Name   account_number   day   transaction_number  other data(sales,
> gross margin, ext)
>
> First I rollup on transaction number getting sums of the other data.
> Then, rollup by name, account_number and day (the combination of name
> and account_number identifies a person and persons might have purchases
> on different days), again getting sums along the way.  This strategy
> requires separate sorts for each rollup.  The final file would look
> something like this:
>
> Name   account_number   sum_of_sales sum_of_gross_margin.....
>

Scott,

As Howard has suggested use class statement in Proc Summary. If that gives
the solution within the time-frame you can tolerate, then it is fine. If you
still want to try out other ways to minimize the running time, then come
back to SAS-L with some simulated data set and the desired output you expect
from such a data set -  then it may be easier for someone on SAS-L to spend
the least time in finding a solution. The variable type/length of the key
variables are at least to be known in advance for a practical solution.

Regards,

Muthia Kachirayan