怎么利用hash或sql将不同观测按要求重新组合？ [推广有奖]

31楼

semenljw

发表于 2013-3-17 10:43:10

经典，呵呵。

32楼

hongxx 发表于 2013-3-22 13:05:50

思路：计算需要重复的记录的重复次数，然后为每一个记录分配到某一个组中。代码看起来简短一些。

data test;
input subjn dt Rx $ x;
cards;
1 1 A 11
1 2 B 12
1 3 C 13
2 3 A 21
2 7 C 22
2 7 B 23
2 12 D 24
3 1 B 31
3 1 C 32
3 5 A 33
3 8 B 34
3 11 B 35
3 11 D 36
3 11 E 37
;
proc sql;
create table test_sorted as
select *, count(*) as rx_types
from test
group by subjn, dt
order by subjn, dt;
quit;
data test_howmany_group;
set test_sorted;
retain howmany_group;
if first.subjn then howmany_group=1;
if first.dt then howmany_group=howmany_group*rx_types;
if last.subjn then output;
by subjn dt;
run;
proc sql;
create table test_howmany_group2 as
select a.*,b.howmany_group
from test_sorted a , test_howmany_group b
where a.subjn=b.subjn
order by subjn,dt;
quit;
data test_goup;
set test_howmany_group2;
retain group_interval;
if first.dt then group_interval=0;
do i=1 to howmany_group/rx_types;
if i=1 then group_interval=group_interval+1;
group=group_interval+rx_types*(i-1); /*将记录输出复制howmany_group/rx_types次，为每次复制赋值group标志*/
output;
end;
by subjn dt;
drop i;
run;
proc sort data=test_goup(drop=rx_types group_interval howmany_group) out=test_goup_final;
by subjn group dt;
run;