|Date: ||Tue, 29 Oct 2002 16:25:39 -0500|
|Sender: ||"SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>|
|Subject: ||Re: SAS Report question|
I don't think it matters how you get the data into SAS. Whatever the method,
you want to do it only once. And when you have a 200K x 100 = 20M SAS
dataset, you only want to hit that once (unless you can cheaply index on the
Two basic approaches:
1. Split the dataset into 100 smaller datasets and then use a macro-based
series of reporting steps for each market.
2. Run the report generators against the big (all markets) dataset, using BY
processing, capture the output on disk, then post-process to reorder as
But #2 is kind of old-fashioned. I did a coder's corner paper on it years
Today, output can be delivered via HTML, so the sequence in which the user
views it is flexible, and also untied from the sequence in which SAS
A little more information would help. What is your delivery medium? Paper?
HTML screen? What SAS procedure(s) will generate the reports?
On Mon, 28 Oct 2002 07:27:37 -0500, Hayes, Jim M <Jim.M.Hayes@DELTA.COM>
>I am producing a set of 18 reports for a client by extracting data from a
>Teradata warehouse using BTEQ/SAS under MVS (I know PROC SQL would be a
>better way to go, but we do not have the SAS/ACCESS Teradata product in
>house). The reports have different views of the same set of data and will
>be keyed by market (i.e. London, Paris, etc.).
>My client wants to all 18 reports for each market to be output together
>(i.e. the overall London report, followed by the London subreports,
>by the overall Paris report and all of its subreports, etc.).
>I'm thinking this can be done in a macro, but am not sure of the best
>approach. I've considered creating a set of macros which selects records
>based on the market then running the resultant extract through each report
>but this seems to be a pretty processing intense way of doing the job
>that there will be 100+ markets to be reported on each of which will have
>some 200,000 input records per month and the reports contain 2 and 24
>worth of data). Any suggestion on how this particular cat can be skinned
>Delta Airlines, Network Analysis