Date: Thu, 19 Jun 2003 18:19:46 -0400
Reply-To: Richard Graham <richardwgraham@earthlink.net>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Richard Graham <richardwgraham@EARTHLINK.NET>
Organization: Glencairn Consulting Group
Subject: Re: SAS Loop Logic Question
Content-Type: text/plain; charset="iso-8859-1"
Here's one approach......
/*Get the contents of the data set*/
proc contents data = wa.integral_sales out=work.contents(keep=name varnum)
noprint;
run;
/* you will probably want to exclude your identifying variables from the
contents */
proc sort data=work.contents;
where name not in ('id1' 'id2' etc);
by varnum;
run;
/*This next data step creates macro variables from the contents data set and
counts how many there are */
data _null_;
set work.contents end=eof;
call symput('v'||trim(left(put(_n_,best.))),name);
if eof then call symput('totv',trim(left(put(_n_,best.))));
run;
/*This step allows you to reference the macro variables (indirect variable
references) in the data step */
%macro test;
data libname.dsname_a;
set libname.dsname;
array vars(&totv) %do _i = 1 %to &totv; &&v&_i %end;;
.....
run;
%mend test;
%test;
HTH
Richard Graham
Vice President
Glencairn Counsulting Group, Inc.
----- Original Message -----
From: "STR" <setar@MINDSPRING.COM>
To: <SAS-L@LISTSERV.VT.EDU>
Sent: Thursday, June 19, 2003 5:35 PM
Subject: Re: SAS Loop Logic Question
> I may have over simplified my dataset for the purposes of this
> discussion. Actually, there are approx. 1000 variables I need to
> analye for each observation. Unfortunately, the variables are not
> numbered in succession. They are unique 5 digit numbers. When
> initializing my array, is there an easier solution outside of listing
> all of my variables? Could I make a dataset containing the variable
> names only and somewhow use that as input to the initialiation of the
> array? Of course it sounds logical, but I wouldn't know how to tackle
> this.
>
> I am greatly for the expertise you may offer.
>
> rjf2@CDC.GOV (Fehd, Ronald J. PHPPO) wrote in message
news:<9D17D648E4EBD311AD45009027D0DF9305002BBF@MCDC-ATL-64>...
> > > From: STR [mailto:setar@MINDSPRING.COM]
> > > I am in need of devising a loop to perform the following
> > > given the sample dataset below:
> > > yr1 yr2 yr3
> > >
> > > EMP1 10 5 6
> > > EMP2 2 7 5
> > > EMP3 3 3 3
> > >
> > > (I have simplefied the dataset for practical purposes)
> > >
> > > For each EMP, I want to add up the ratings for all 3 years.
> > > If the sum of all 3 years for that EMP is greater than 10,
> > > then I want to write out the highest value to another data
> > > set. Then I want perform the same operation, eliminating the
> > > highest value identified in the previous step from the
> > > calculation. I want to continue this loop until the
> > > remaining values left to calculate aggregate to a value less than 10.
> >
> > you are in need of a major RTFM session.
> > the functions you are looking for are
> > sum(of Years(*))
> > and similarly:
> > max(of Years(*))
> >
> > based on your simplified data set structure
> > you'll need to define an array:
> > array Year(3);
> > if your year variables are not contiguous in the data set
> > or have different prefixes than Year
> > then you will have to list them:
> >
> > DATA LESS-THAN-10
> > EXACTLY-10
> > MORE-THAN-10
> > MAX (keep = Max Fred Georgette)
> > ;
> > array Year(3) year1 yr2 yer3;
> > drop I;
> >
> > something like:
> > SumYears = sum(of Years(*));
> > if SumYears gt 10 then output MORE-THAN-10;
> > Max = max(of Years(*));
> > if Max gt 10 then output MAX;
> > do I = 1 to dim(Years);*...;
> >
> > that should get you going.
> >
> > Ron Fehd the macro maven CDC Atlanta GA USA RJF2@cdc.gov
> >
> > RTFM: I'm an engineer, I don't get paid to know,
> > I get paid to know where to look it up.
> > RTFM: Read The Finite Manual.
> >
> > RTFM: I'm The Maven I don't get paid to know,
> > I volunteer to tell other people what to look up.
> > RTFM: Read The Finite Manual.
>
