Date: Tue, 23 May 2006 22:24:27 -0700
Reply-To: David L Cassell <davidlcassell@MSN.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: David L Cassell <davidlcassell@MSN.COM>
Subject: Re: Creating combinations in SAS
Content-Type: text/plain; format=flowed
saurabhkdas@GMAIL.COM wrote back:
>Thank you all for replying , and sorry I am late, because of these
>faking ISP s who promise something and deliver something else ( yes my
>Internet connection ... its pathetic) . Yes I am sorry that I am new
>and asking too many questions .... and I really dont want fully baked
>bread here ....all I would need is direction and yes a little help
>would help me get a grip on the learning curve. Alrite here we go :-
>libname temp 'D:\saurabh\APRs\NDD Correlation Update\';
>PROC IMPORT DATAFILE= "D:\saurabh\APRs\NDD Correlation
> OUT= temp.NddAgent
> DBMS=EXCEL2000 REPLACE;
>proc corr data = temp.NDDAgent outp =
>temp.CorrNDDAgent(type=Corr)noprint NOSIMPLE ;
>PROC EXPORT DATA= temp.CorrNDDAgent
> OUTFILE= "D:\saurabh\APRs\NDD Correlation
> DBMS=EXCEL2000 REPLACE;
>PROC REG data = temp.CorrNDDAgent OUTEST=temp.CorrNDDAgentout;
>model NowDuepersystime =
>That reg part .. I am doing it manually. I would need to learn and get
>it done through a macro .. as everyone suggested . I would try to get
>the combination done too .. so that I would not need to depend on
>anything else ..than SAS... taking 4 variable at a time is what we can
>take as of now as suggested by my quality lead. I need to struggle else
>I will not learn. But I require your help too gentlemen. My next step
>would be getting the combination done and having the macro done for
>proc reg . Would pick up all the combination and run reg on it all I
>need is the output into a data set from all combination. If you think I
>am struggling.. I am :( . I ll get back once I move to the next step
>or not ... could be in the next minute or the next hour ... sorry folks
Okay, let me make a few comments.
Running the data through PROC CORR and then running regressions on
the correlation coefficients seems like a really bad idea. I don't see any
justification for this.
And doing so over and over again, picking out the maximum adjusted
R-squared like it would be meaningful here, is just asking for problems.
Why are you doing this at all? What is the larger goal? Explain what you
are *really* trying to do, and why, and how the above work is supposed
to achieve those goals.
Don't start on a macro. Right now you have nothing worth macro-izing.
Sorry to be so blunt, but this is looking even worse than when you
started explaining it to us. Write back to SAS-L and please answer my
David L. Cassell
3115 NW Norwood Pl.
Corvallis OR 97330
Express yourself instantly with MSN Messenger! Download today - it's FREE!