Proc Means loop Macro with interactions

[Shmuel]

Final code, limited up to 3 factors - you are invited to run it :

%macro meansANOVA(data, totalvars);
     /* loop per variable * per factor combination */
       data _null_;
           class1 = translate("&class",'_',' ');
           call symputx('class1',class1);
     run;
    %do i = 1 %to &totalvars;
            proc means data = &data clm mean std stderr;
                    class &class;
                    var m&i;
                   ods out summary=&data._&class1._Means&i;
    %end;
%mend meansANOVA;

%macro class_f(factors, nvars, dsn);
       /* calculate class = combination of factors and run meansANOVA */
      %let nf = %sysfunc(countw(&factors));
            /* single factor */
     %do i=1 %do &nf;
             %let f&i = %scan(&factors, &i);

             %let class = &&f&i;
             %meansANOVA(&dsn, &nvars);
     %end;
           /* factors combination */
     %if %eval(&nf) = 3 %then %do;
            %let class = &f1 &f2; %meansANOVA(&dsn, &nvars);
            %let class = &f1 &f3; %meansANOVA(&dsn, &nvars);
            %let class = &f2 &f3; %meansANOVA(&dsn, &nvars);
    %end; %else
   %if %eval(&nf) = 2 %then %do;
           %let class = &f1 &f2;  %meansANOVA(&dsn, &nvars);
   %end;
%mend class_f;

to run the code use:

     %class_f(factors, nvars, dsn);

where factors is list of factor variable names seperated by space.

            nvars   is total number of variables m1 - m&nvars.

            dsn      is the input dataset name.

I could not run it  and hpe you can debug it in case of need.

[Tom]

Did you try using the OUTTABLE option on PROC UNIVARIATE instead?

data have ;
  length sample $20 id 8 factor1 factor2 $8 m1-m5 8;
  input SAMPLE ID Factor1 Factor2 m1 m2 m3 m4 m5 ;
cards;
NCAT-00121-LC-01 1 S1F2 S2F2 1.1242 0.9108 1.2078 0.9698 1.3997
NCAT-00122-LC-01 2 S1F2 S2F2 0.9238 1.1207 1.3090 1.1214 1.1122
NCAT-00123-LC-01 3 S1F2 S2F2 0.9008 1.0764 1.0655 1.0333 0.9139
NCAT-00125-LC-01 4 S1F2 S2F2 0.9580 1.0099 0.9423 0.7857 0.7829
NCAT-00126-LC-01 5 S1F2 S2F2 0.9911 1.1804 1.1002 0.8962 0.7059
NCAT-00127-LC-01 6 S1F2 S2F2 0.9337 0.7790 0.9809 0.7370 1.1507
NCAT-00128-LC-01 7 S1F2 S2F2 0.8911 0.8306 0.9966 0.9008 1.1129
NCAT-00129-LC-01 8 S1F2 S2F2 0.7743 0.9075 0.9006 0.7545 1.0823
NCAT-00130-LC-01 9 S1F2 S2F2 0.8778 1.2467 1.1428 1.0346 1.1459
NCAT-00131-LC-01 10 S1F2 S2F2 1.1837 0.9635 1.0022 0.9484 1.2579
NCAT-00132-LC-01 11 S1F2 S2F2 1.0674 0.8744 1.1192 1.3258 1.1429
NCAT-00133-LC-01 12 S1F2 S2F2 0.9934 0.9600 0.9988 1.0292 1.1051
NCAT-00134-LC-01 13 S1F2 S2F2 1.2496 1.1706 1.2041 1.2644 1.2787
NCAT-00136-LC-01 14 S1F2 S2F1 0.8153 0.8933 1.3870 1.1998 1.3597
NCAT-00137-LC-01 15 S1F2 S2F1 1.1740 0.9647 1.2137 1.0219 0.9354
NCAT-00138-LC-01 16 S1F2 S2F1 1.0085 0.9042 1.0351 1.0457 0.9850
NCAT-00139-LC-01 17 S1F2 S2F1 0.7983 0.9432 1.0341 1.1845 1.1765
NCAT-00140-LC-01 18 S1F2 S2F1 0.7831 0.9264 0.9654 1.0773 1.1253
NCAT-00141-LC-01 19 S1F2 S2F1 1.1816 1.0947 1.2754 0.9486 1.5532
NCAT-00142-LC-01 20 S1F2 S2F1 0.9824 0.8790 1.0780 0.6967 0.8854
NCAT-00143-LC-01 21 S1F2 S2F1 0.7157 0.8337 0.9973 0.8521 1.0414
NCAT-00144-LC-01 22 S1F2 S2F1 0.8874 0.8590 1.1573 0.8433 1.3164
NCAT-00145-LC-01 23 S1F2 S2F1 0.9024 0.8867 1.0910 1.0155 1.2733
NCAT-00146-LC-01 24 S1F2 S2F1 1.2156 0.8162 1.1029 1.1553 1.2157
NCAT-00147-LC-01 25 S1F2 S2F1 0.9392 1.0253 1.2494 1.2901 1.6068
NCAT-00148-LC-01 26 S1F2 S2F1 0.9844 0.8720 1.1468 0.9804 0.9447
NCAT-00149-LC-01 27 S1F2 S2F1 0.8876 1.1102 1.0676 1.1175 1.0071
NCAT-00150-LC-01 28 S1F2 S2F1 1.0687 0.9460 1.0729 1.2188 1.0361
;
proc univariate data=have outtable=want noprint;
 class factor1 factor2 ;
 var m1-m5 ;
run;

Here are some of the statistics that it produces.

Obs    _VAR_    factor1    factor2    _NOBS_     _MIN_     _MAX_     _STD_

  1     m1       S1F2       S2F1        15      0.7157    1.2156    0.15297
  2     m1       S1F2       S2F2        13      0.7743    1.2496    0.13365
  3     m2       S1F2       S2F1        15      0.8162    1.1102    0.08760
  4     m2       S1F2       S2F2        13      0.7790    1.2467    0.14597
  5     m3       S1F2       S2F1        15      0.9654    1.3870    0.11468
  6     m3       S1F2       S2F2        13      0.9006    1.3090    0.11919
  7     m4       S1F2       S2F1        15      0.6967    1.2901    0.16173
  8     m4       S1F2       S2F2        13      0.7370    1.3258    0.18038
  9     m5       S1F2       S2F1        15      0.8854    1.6068    0.22222
 10     m5       S1F2       S2F2        13      0.7059    1.3997    0.19223

[Reeza]

Use STACKODS with proc means. And look at CLASSDATA to build custom class levels.