Date: Thu, 2 Dec 2004 08:16:49 -0500
Reply-To: Ed Heaton <EdHeaton@WESTAT.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Ed Heaton <EdHeaton@WESTAT.COM>
Subject: Re: Calculating with macro values (ie comparing to a # in if/the
n statement)
Content-Type: text/plain
Venita,
I'm having a slightly hard time following your code. Is this what you want
to do? I have changed the %IF statements to IF statements because I don't
see the need for them.
%do i=1 %to &nVars ;
Proc reg data=out_hetero ;
Model res2 = &&var&i &&var&i.._sq ;
Ods output parameterEstimates=var_est&i ;
Run ;
Proc print data=var_est&i ;
Run ;
Data _null_ ;
Set var_est&i end=endOfFile ;
If endOfFile ;
If (
( ProbT gt &alpha)
or missing(ProbT)
) then call execute(
"Proc reg data=out_hetero ;"
|| " Model res2 = &&var&i ;"
|| " Ods output parameterEstimates=var_est&i ;"
|| "Run;"
) ;
Run ;
Proc print data=var_est&i ;
Run ;
Data var_est_&i ;
Length Variable $20 ;
Set var_est&i end=endOfFile ;
Where ( variable ne 'Intercept') ;
If (
endOfFile
& ( ( ProbT lt 0 ) or (ProbT gt &alpha) )
) then call execute("Data var_est_&i ; Set _null_ ; Run ;") ;
Run ;
Data hetero_var ;
Set
hetero_var
var_est_&i
;
Run ;
%end ;
Proc print data=hetero_var ;
Run ;
Ed
Edward Heaton, SAS Senior Systems Analyst,
Westat (An Employee-Owned Research Corporation),
1600 Research Boulevard, RW-3541, Rockville, MD 20850-3195
Voice: (301) 610-4818 Fax: (301) 610-5128
mailto:EdHeaton@Westat.com http://www.Westat.com
-----Original Message-----
From: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU] On Behalf Of DePuy,
Venita
Sent: Wednesday, December 01, 2004 12:27 PM
To: SAS-L@LISTSERV.UGA.EDU
Subject: Re: Calculating with macro values (ie comparing to a # in if/the n
statement)
Hi again -
Thanks for all the responses earlier!
And you're right, in most cases I was complicating things I didn't need to.
But I have one case (so far) where I'm having trouble . .
Background: proc reg, model y = x x^2; if x^2 isn't significant, model
y=x; if x isn't significant, set the output file to _null_. It's a method
of iteratively looking at heteroscedasticity by regressing residuals from an
earlier model. (Yes, I'm sure this is the way I want to do it).
So take the output from the y=x x^2 model, look at the last line, if Probt>
&alpha or Probt=. then run the next model;
if that model has Probt>&alpha or Probt=. then set that output file =null,
since the next step is stacking these outputs into a file w/ data & set
statements.
For some reason, my first comparison is working but my second isn't, if that
makes sense.
In this case trt (a 0/1 variable) is significant, but trt_sq isn't (Probt=.
as a 0/1 variable squared is redundant), and trt2 and trt2_sq are not either
significant (p's .16,.20)
Code, output, and log follow.
Thanks for the help!
Here's the section of code:
%do i = 1 %to &nvars;
proc reg data=out_hetero ;
model res2 = &&var&i &&var&i.._sq;
ods output ParameterEstimates = var_est&i; run;
proc print data=var_est&i;
data _null_;
set var_est&i end=ENDOFFILE;
if not ENDOFFILE then delete;
%if ((Probt > &alpha) OR (Probt= . )) %then %do;
call execute (" proc reg data=out_hetero ; model res2 =
&&var&i ;
ods output
ParameterEstimates = var_est&i;run;");
run;%end;run;
proc print data=var_est&i;
data var_est_&i;
length Variable $20; set var_est&i end=ENDOFFILE;
where variable ne 'Intercept';
%if (ENDOFFILE =1) and ((Probt > &alpha) OR (Probt < 0 ) ) %then
%do;
call execute ("data var_est_&i; set _null_;run;"); %end;
data hetero_var; set hetero_var var_est_&i; run;
proc print data=hetero_var;
Here's the section of output (print sections only as others would be
redundant):
Obs Model Dependent Variable DF Estimate StdErr tValue
Probt
1 MODEL1 res2 Intercept 1 0.01052 0.00152 6.94
<.0001
2 MODEL1 res2 trt B 0.01033 0.00303 3.41
0.0008
3 MODEL1 res2 trt_sq 0 0 . . .
Obs Model Dependent Variable DF Estimate StdErr tValue
Probt
1 MODEL1 res2 Intercept 1 0.01052 0.00152 6.94
<.0001
2 MODEL1 res2 trt 1 0.01033 0.00303 3.41
0.0008
HETERO_VAR 10:07 Wednesday, December 1,
2004
Obs Variable Model Dependent DF Estimate StdErr tValue
Probt
1 trt MODEL1 res2 1 0.01033 0.00303 3.41
0.0008
Estimating variance function heteroscedasticity for: trt2
Obs Model Dependent Variable DF Estimate StdErr tValue
Probt
1 MODEL1 res2 Intercept 1 0.01355 0.00166 8.16
<.0001
2 MODEL1 res2 trt2 1 0.00299 0.00160 1.87
0.0633
3 MODEL1 res2 trt2_sq 1 -0.00134 0.00104 -1.28
0.2005
Obs Model Dependent Variable DF Estimate StdErr tValue
Probt
1 MODEL1 res2 Intercept 1 0.01245 0.00142 8.75
<.0001
2 MODEL1 res2 trt2 1 0.00187 0.00135 1.39
0.1655
HETERO_VAR
Obs Variable Model Dependent DF Estimate StdErr tValue
Probt
1 trt MODEL1 res2 1 0.01033 0.00303 3.41
0.0008
2 trt2 MODEL1 res2 1 0.00187 0.00135 1.39
0.1655
Here's the section of log (tried to remove nonpertinent sections):
MLOGIC(OLS): %DO loop beginning; index variable I; start value is 1; stop
value is 2; by value is 1.
NOTE: The data set WORK.VAR_EST1 has 3 observations and 8 variables.
NOTE: PROCEDURE REG used (Total process time):
NOTE: There were 3 observations read from the data set WORK.VAR_EST1.
NOTE: PROCEDURE PRINT used (Total process time):
MLOGIC(OLS): %IF condition ((Probt > &alpha) OR (Probt= . )) is TRUE
NOTE: There were 3 observations read from the data set WORK.VAR_EST1.
NOTE: DATA statement used (Total process time):
NOTE: CALL EXECUTE generated line.
1 + proc reg data=out_hetero ; model res2 = trt
;
odsoutput ParameterEstimates = var_est1;run;
NOTE: The data set WORK.VAR_EST1 has 2 observations and 8 variables.
NOTE: PROCEDURE REG used (Total process time):
NOTE: There were 2 observations read from the data set WORK.VAR_EST1.
NOTE: PROCEDURE PRINT used (Total process time):
MLOGIC(OLS): %IF condition (ENDOFFILE =1) & ((Probt > &alpha) OR (Probt <
0)) is FALSE
NOTE: There were 1 observations read from the data set WORK.VAR_EST1.
WHERE variable not = 'Intercept';
NOTE: The data set WORK.VAR_EST_1 has 1 observations and 8 variables.
NOTE: DATA statement used (Total process time):
NOTE: There were 0 observations read from the data set WORK.HETERO_VAR.
NOTE: There were 1 observations read from the data set WORK.VAR_EST_1.
NOTE: The data set WORK.HETERO_VAR has 1 observations and 8 variables.
NOTE: DATA statement used (Total process time):
NOTE: There were 1 observations read from the data set WORK.HETERO_VAR.
NOTE: PROCEDURE PRINT used (Total process time):
MLOGIC(OLS): %DO loop index variable I is now 2; loop will iterate again.
NOTE: The data set WORK.VAR_EST2 has 3 observations and 8 variables.
NOTE: PROCEDURE REG used (Total process time):
NOTE: There were 3 observations read from the data set WORK.VAR_EST2.
NOTE: PROCEDURE PRINT used (Total process time):
MLOGIC(OLS): %IF condition ((Probt > &alpha) OR (Probt= . )) is TRUE
NOTE: There were 3 observations read from the data set WORK.VAR_EST2.
NOTE: DATA statement used (Total process time):
NOTE: CALL EXECUTE generated line.
1 + proc reg data=out_hetero ; model res2 = trt2 ;
ods output ParameterEstimates = var_est2;run;
NOTE: The data set WORK.VAR_EST2 has 2 observations and 8 variables.
NOTE: PROCEDURE REG used (Total process time):
NOTE: There were 2 observations read from the data set WORK.VAR_EST2.
NOTE: PROCEDURE PRINT used (Total process time):
MLOGIC(OLS): %IF condition (ENDOFFILE =1) and ((Probt > &alpha) OR (Probt <
0 )) is FALSE
NOTE: There were 1 observations read from the data set WORK.VAR_EST2.
WHERE variable not = 'Intercept';
NOTE: The data set WORK.VAR_EST_2 has 1 observations and 8 variables.
NOTE: DATA statement used (Total process time):
NOTE: There were 1 observations read from the data set WORK.HETERO_VAR.
NOTE: There were 1 observations read from the data set WORK.VAR_EST_2.
NOTE: The data set WORK.HETERO_VAR has 2 observations and 8 variables.
NOTE: DATA statement used (Total process time):
NOTE: There were 2 observations read from the data set WORK.HETERO_VAR.
NOTE: PROCEDURE PRINT used (Total process time):
MLOGIC(OLS): %DO loop index variable I is now 3; loop will not iterate
again.
NOTE: There were 2 observations read from the data set WORK.HETERO_VAR.
NOTE: DATA statement used (Total process time):
