Data-set A: n type x0 x1 groups track block0 1 0 var1 value1 1 1 1 1 0 var2 value2 1 2 1 1 0 var3 value3 1 3 1 1 0 var4 value4 1 4 1 1 0 var5 value5 1 5 1 1 0 var6 value6 1 6 1 1 0 var7 value7 1 7 1 2 0 var1 value1 2 1 1 2 0 var2 value2 2 2 1 2 0 var3 value3 2 3 1 2 0 var4 value4 2 4 1 2 0 var5 value5 2 5 1 2 0 var6 value6 2 6 1 2 0 var7 value7 2 7 1 2 0 var4 value4 2 8 2 2 0 var5 value5 2 9 2 2 0 var6 value6 2 10 2 2 0 var7 value7 2 11 2 Data-set B: n type groups var1 var2 var3 var4 var5 var6 var7 1 0 1 value1 value2 value3 value4 value5 value6 value7 2 0 2 value1 value2 value3 value4 value5 value6 value7 2 0 2 value4 value5 value6 value7 Hello, so basically, I want to take data-set A and turn it into data-set B. I assigned groups, track, and block0 variables to try and help me separate, transpose, and stack this data, which i attempt with the macro below. This is just a sample, of course. I have block0 that go up to around 120 and groups that go up to 20. I also have many different types. I have been using the code below: %macro separate;
%do i=1 %to 19;
data d&i.; set x9;
if groups=&i.;run;
%let max=;
data _null_; set d&i.(obs=1);
call symput('max',groups);run;
%do t=1 %to &max.;
data e&t.; set d&i.;
if block0 = &t.;run;
proc sort data=e&t.; by n type groups;
proc transpose data=e&t. out=f&t.(drop=_name_); by n type groups; var value; id x0;
proc append data=f&t. force base=xxx&i.;
%end;
proc append data=xxx&i. force base=final;
%end;
%mend;
%separate; This produces the output I want, for a small subset, I think.... but I believe it to be inefficient and not robust. Is there a better way to do this? Please note, that var1-var7 will always be the same and equal, but the frequency of how var4-var7 repeat per n varies, var1-var3 will only occur once per n. Value1-Value7 can take on any values..
... View more
