Reply to 按照某个变量的值拆分数据集（为循环和批量分析服务） on Mon, 25 May 2020 01:24:30 GMT

谢远涛 — Mon, 25 May 2020 01:24:30 GMT

/*按照变量date的值来拆分数据集
如date=19960131生成一个数据集，date=19960228生成一个
现在有很多个，到20080731了
*/
/*方法一
思路是先把distinct date的值变为宏变量，然后用data步输出。
反复读数据，会很慢
*/
data a;
do date=1991 to 2011;
output;
end;
run;
data null;
set a end=last;
call symputx('date'||left(n),date);
if last then call symput ('n',n);
run;
%macro split;
%do i=1 %to &n;
data b&i;
set a;
where date=&&date&i;
run;
%end;
%mend;
%split;

/方法二/
data date;
date=input("19960131",yymmdd8.);
output;
date=input("19960228",yymmdd8.);
output;
date=input("19960309",yymmdd8.);
output;
run;

data null;
set date;
call execute ("data data_"||strip(put(date,yymmddn8.))||";set a;if date="||date||";run;");
run;

/方法三/

data a;
format target_day $8.;
input target_day $;
cards;
20120730
20120731
20120810
20120817
;run;

%macro segmentation();
proc sort data = a out = work.t1(keep = target_day) nodupkey;
by target_day;
run;
data null;
set work.t1 nobs=n;
call symputx(compress("set_"||n), put(input(target_day, yymmdd8.), yymmddn8.));
if n = 1 then call symputx('n',n);
run;
data %do ii = 1 %to &n.;
data_&&set_&ii..
%end;;
set a;
select (target_day);
%do ii = 1 %to &n.;
when ("&&set_&ii..") output data_&&set_&ii..;
%end;
otherwise;
end;
run;
proc sql;
drop table work.t1;
quit;
%mend;
%segmentation();