/*方法一*/
DATA t ; merge cm(in=a ) treat ; by subjid ; if a ;
run;
proc sort data=T nodupkey out=t3;
by subjid cmdecod;run;
/*方法二*/
DATA t ; merge cm(in=a ) treat ; by subjid ; if a ;
run;
data t2 ; set t ; by subjid cmdecod;
if first.cmdecod;
run;
run;
data t2 ; set t ; by subjid cmdecod;
if first.cmdecod;
run;
/*方法三*/
proc sql
noprint;
create table cmtosum as
select unique (c.cmdecod) as cmdecod, c.subjid, t.trtcd
from cm as c, treat as t
where c.subjid = t.subjid
order by subjid, cmdecod;
quit;
/*依組別(藥名)，把資料檔由直向拉成橫向*/
/*方法一*/
Proc sort data = counts; by cmdecod trtcd; run;
Data cc2 ; set counts; by cmdecod trtcd;
/*記住3個變項值，不然數據長成一斜線*/
Retain n1-n3;
/*新組別時，先清空數據*/
Array nn {3} n1-n3;
if first.cmdecod then do i =1 to 3;
nn{i}=.;
end;
/*指定數值，亦可用Array的方式 nn{trtcd} ，但本例 trtcd有missing，固改用 if then 的方式*/
if trtcd=. then n1=frequency;
if trtcd=0 then n2=frequency;
if trtcd=1 then n3=frequency;
/*輸出每組的最後一筆*/
if last.cmdecod;
run;
/*方法二*/
Proc sort data = counts; by cmdecod trtcd; run;
/*利用 Merge ，分組別進行*/
Data cm;
merge counts(where = (trtcd = 1) rename = (frequency = count1))
counts(where = (trtcd = 0) rename = (frequency = count2))
counts(where = (trtcd = .) rename = (frequency = count3))
end = eof;
by cmdecod;
run;
/*方法三*/
/*較不建議，若是 ID 的變項值為 missing ，資料會被刪除*/
Proc transpose data=counts prefix=nn out=cc ;
by cmdecod ; id trtcd;
var Frequency;
run;
/*方法三*/
/*較不建議，若是 ID 的變項值為 missing ，資料會被刪除*/
Proc transpose data=counts prefix=nn out=cc ;
by cmdecod ; id trtcd;
var Frequency;
run;
部分程式引用，Ref: Jack Shostak, SAS Programming in the Pharmaceutical Industry.
