Date:         Fri, 7 Jan 2000 12:53:05 +0000
Reply-To:     Peter Crawford <Peter@CRAWFORDSOFTWARE.DEMON.CO.UK>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Peter Crawford <Peter@CRAWFORDSOFTWARE.DEMON.CO.UK>
Organization: Crawford Software Consultancy Limited
Subject:      Re: challenging problem

Sounds like a job for SAS/AF, if interaction with a user is necessary,
but not otherwise.

One could use call execute() to generate the code
Assume you have a macro %foremacr needing parameters to generate the
required forecasts from variables of the merged data.

data shortlst;
  merge data.one( in=one )
        dat2.two( in=two );
     by whatever;
  if one and two then delete;
  call execute(  '%foremacr(' );
  call execute(  'parm1=' || data variables as necessary );
  call execute( ',parm2=' || data variables as necessary );
  call execute( ',parm3=' || data variables as necessary );

  call execute( ',parmN=' || data variables as necessary );
  call execute( ' );   )  );
run;

The call execute() above is stylised to give the idea.
I hope it becomes obvious that call execute() just generates code from
run-time values in variables and constants.

Hope this helps



Gijs Noordhoek <gijs.noordhoek@iname.com> writes
>Here's a challenge for all SAS expert out there:
>
>I'm merging two datasets containing new data, provided by two different
>companies. There's a partial overlap in the two datasets, and when
>overlap occurs, several checks have to be done to be able to choose the
>figures from one of the datasets as the actual data.
>
>One possible check is to see whether the values compare with the figure
>obtained by a lineair forecast from past data.
>
>The problem is: I cannot create a forecast from within the datastep.
>First of all because I cannot pass datastep variables to a macro from
>within the datastep (because macro variables are set after the 'run'
>command has passed), second of all because you cannot nest datasteps or
>run a proc forecast within a datastep.
>
>Creating all possible forecasts in advance is impossible: I would have
>to create at least 10000 forecasts for each month in the database,
>containing up to 10 years. Quite a waste of processing time when you
>imaging that I will probably only need a dozen of them.
>
>What I would like to be able to do, is to run an independent SAS
>program that I pass a few parameters to. Just as a *function* works in
>any ordinary programming language.
>
>Does anyone know if this is somehow possible and if not, if there's an
>alternative way of solving this problem?
>
>
>Sent via Deja.com http://www.deja.com/
>Before you buy.

--
Peter Crawford