Date:         Fri, 31 Jul 1998 07:31:57 GMT
Reply-To:     quenzer@MY-DEJANEWS.COM
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         quenzer@MY-DEJANEWS.COM
Organization: Deja News - The Leader in Internet Discussion
Subject:      Re: Restate min and max problems

In article <00092BE2.3171@medstat.com>,
  Peter Chen <Peter_Chen@MEDSTAT.COM> wrote:
>      Our AF application will list the name, type, minimum, maximum, and
>      percentage of missing values for BOTH numeric and char variables,
>      and will also list average and total sum for numeric variables only.
>
>      Our datasets can have 40,000,000 obs and 300 vars.
>
>      The problem: The app is running too slowly. It may take a half day to
>      run the job on a dataset of such a size mentioned above.
>
>      Thanks for any suggestions.
>
>      Peter
>
Hi Peter
 
I think your problem is that you read each record in the af-program.
If you would use a data step it might be much faster.
But then the data step must be generated at run time.
 
So how about that:
- Get the field description in the af-program
- generate a data step and submit it. In this data step the the file
  is analyzed and the informations are stored in a collective file
- at last open the collective file and generate your report
 
Hope this helps
Michael
 
-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/rg_mkgrp.xp   Create Your Own Free Member Forum