Date:         Wed, 2 May 2001 15:43:00 -0400
Reply-To:     "Oltsik, Myra" <moltsik@RESPONSE.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Oltsik, Myra" <moltsik@RESPONSE.COM>
Subject:      Re: simple macro question - %let
Comments: To: Ian Whitlock <whitloi1@WESTAT.COM>
Content-Type: text/plain; charset="iso-8859-1"

Wow! There's still so much to learn and so many ways of doing things. Thanks
IAN.

Myra

                -----Original Message-----
                From:   Ian Whitlock [mailto:whitloi1@WESTAT.COM]
                Sent:   Wednesday, May 02, 2001 3:24 PM
                To:     SAS-L@LISTSERV.UGA.EDU; Myra Oltsik
                Cc:     Ian Whitlock
                Subject:        Re: simple macro question - %let

                Myra,

                Assuming your step

                   DATA _NULL_;
                      SET STATES;
                      DO I = 1 TO &OBSV;
                         IF _N_ = I THEN DO;
                            CALL SYMPUT('STATE'||LEFT(PUT(I,2.)),ST);
                         END;
                      END;
                   RUN;

                should have been

                   DATA _NULL_;
                      DO I = 1 TO &OBSV;
                         SET STATES;
                         IF _N_ = I THEN DO;
                            CALL SYMPUT('STATE'||LEFT(PUT(I,2.)),ST);
                         END;
                      END;
                   RUN;

                you could simplify the code to

                   PROC SUMMARY DATA=NCLM.EARNPQ MISSING NWAY;
                      VAR SEARNED;
                      CLASS MASTERST;
                      WHERE MASTERST NE 'NY';
                      FORMAT SEARNED DOLLAR20.2;
                      OUTPUT OUT=STATES(RENAME=(MASTERST=ST) DROP=_TYPE_
_FREQ_) SUM=;
                   RUN;

                   DATA _NULL_ ;
                      SET STATES(KEEP=ST) ;
                      CALL EXECUTE ( '%MSTCLM(&Q,ST=' || ST || ')' );
                   RUN ;

                Incidentally, CALL EXECUTE also provides a second solution
to the original
                question.  PROC SQL provides a third alternative.  These
tools often
                eliminate the need for arrays of macro variables.  Dare I
say, making them
                almost obsolete?

                Ian Whitlock

                On Wed, 2 May 2001 14:00:33 -0400, Myra Oltsik
<moltsik@RESPONSE.COM> wrote:

                >On Wed, 2 May 2001 10:22:31 -0400, Dan Fiscus
<fiscus@AL.UMCES.EDU> wrote:
                >
                >>Hi,
                >>
                >>This must be an easy one, but as fairly early on in
getting the
                >>macro language, I am stumped...
                >>
                >>I want to store a dataset variable value into a macro
variable.
                >>I have a dataset variable called x and want to store the
current
                >>observation's value of x into the macro variable chosenx.
But
                >>if I say (from inside a data step):
                >>
                >>%let chosenx = x ;
                >>
                >>the value of chosenx becomes the string 'x'.
                >>
                >>Thanks for any suggestions...
                >>
                >>Dan Fiscus
                >>
                >>
                >>--
                >>
                >>Dan Fiscus
                >>Ecologist/Research Assistant             email:
fiscus@al.umces.edu
                >>Appalachian Laboratory                   phone:
301-689-7121
                >>University of Maryland C.E.S.            fax: 301-689-7200
                >>301 Braddock Road
http://www.al.umces.edu
                >>Frostburg, MD 21532   USA
                >
                >Dan,
                >
                >It's not exactly easy. Here's some code I use to do what I
think you want
                >to do: [I have comments below.]
                >
                >PROC SUMMARY DATA=NCLM.EARNPQ MISSING NWAY;
                >   VAR SEARNED;
                >   CLASS MASTERST;
                >   WHERE MASTERST NE 'NY';
                >   FORMAT SEARNED DOLLAR20.2;
                >   OUTPUT OUT=STATES(RENAME=(MASTERST=ST) DROP=_TYPE_
_FREQ_) SUM=;
                >RUN;
                >
                >DATA _NULL_;
                >   SET STATES(KEEP=ST) NOBS=NO;
                >   CALL SYMPUT('OBSV',LEFT(PUT(NO,8.)));
                >   STOP;
                >RUN;
                >
                >%GLOBAL OBSV;
                >
                >DATA _NULL_;
                >   SET STATES;
                >   DO I = 1 TO &OBSV;
                >      IF _N_ = I THEN DO;
                >         CALL SYMPUT('STATE'||LEFT(PUT(I,2.)),ST);
                >      END;
                >   END;
                >RUN;
                >
                >%MACRO MRESVD;
                >
                >   %GLOBAL Z;
                >
                >   %DO Z = 1 %TO &OBSV;
                >      %MSTCLM(&Q,ST=&&STATE&Z);
                >   %END;
                >
                >%MEND MRESVD;
                >
                >%MRESVD;
                >
                >==========
                >
                >On it's first pass, &&STATE&Z resolves to &STATE1. On the
next pass,
                >&STATE1 response to AZ (for Arizona). I learned this in Art
Capenter's
                >Macro class about 1-½ years ago. I don't have the code
memorized, but
                >always available when I need it.
                >
                >Myra