请教：利用nodupkey可以去重复数据，但有没有简单办法把重复数据挑出来

shiyiming

支用将表读入hash中，然后调用find函数，输出，效率还高，不过数据量不能超过其内存大小。

shiyiming

模糊记得可以直接sort输出，刚查了下help，确认了下可以这样（V9.3）

[code:2iomnahs]data abs;
input blh $12. ksdm : $15. rq yymmdd10.;
attrib rq format=yymmdd10.;
cards;
PID000529036 8楼儿童病区 2012-3-13
PID000529036 ICU 2012-3-13
PID000529036 ICU 2012-3-14
PID000529036 ICU 2012-3-15
PID000529036 6楼儿童病区2 2012-3-13
PID000531543 8楼儿童病区 2012-3-14
PID000531543 ICU 2012-3-15
PID000531543 ICU 2012-3-16
;
run;

* V9.3
proc sort data=abs out=_a1 UNIQUEOUT=_a2 NOUNIQUEkey;
        by  ksdm rq;
run;

* V9.1的时候
proc sort data=abs out=_b1 dupout=_b2 nodupkey;
        by ksdm rq;
run;
[/code:2iomnahs]

shiyiming

Good idea! This could be a little more concise:

proc sql;
        create table one_a as
        select *, monotonic() as id
        from abs
        order by ksdm,rq;

        create table two_a as
        select *
        from one_a
        group by ksdm,rq
        having max(id) ne min(id);

        alter table two_a drop id;
quit;

[quote="MerlinZHOU":2pm7kenj][code:2pm7kenj]
data abs;
input blh $12. ksdm : $ 15. rq yymmdd10.;
cards;
PID000529036 8楼儿童病区 2012-3-13
PID000529036 ICU 2012-3-13
PID000529036 ICU 2012-3-14
PID000529036 ICU 2012-3-15
PID000529036 6楼儿童病区2 2012-3-13
PID000531543 8楼儿童病区 2012-3-14
PID000531543 ICU 2012-3-15
PID000531543 ICU 2012-3-16
;

proc sql;

create table one as
select *
from abs
order by
ksdm,rq;

create table two as
select *, monotonic() as id
from one;

create table three as
select *
from two
group by ksdm,rq
having max(id)^=min(id);

alter table three drop id;

quit;
[/code:2pm7kenj][/quote:2pm7kenj]

shiyiming

[quote="faithd":rkoxlgzz]版主的第一种方法是求出相同数据的第一条和最后一条数据,如果要输出所有的重复值,需要修改判断语句如下:
将语句
if first.rq^=last.rq then output two;
修改为:
if first.rq^=last.rq or (last.rq =0 and first.rq=0) then output two;[/quote:rkoxlgzz]


If first.rq and last.rq then delete;