topic Remove less than 10 observations in SAS Procedures

Remove less than 10 observations

q1234 — Thu, 03 Mar 2011 13:22:16 GMT

Hi

Any one can you help with this? I used to use this code to remove less than 10 observations in the same year and industry and it works. But I need to create a dataset for each year and the run the code.

So, can you help me to modify this code to run it for pooled sample including all the years?

Thanks

the dataset
Firm industry year
x 22 2001
z 33 2002
v 33 2002
b 14 2005
w 22 2001
x 22 2002
x 22 2003

The code

proc sort data=data;
by industry;
run;

proc tabulate data=dataset out= dataset_1;
class industry;
table industry*n;
run;

data dataset_1;
set dataset_1;
drop _table_;
drop _page_;
drop _type_;
run;

proc sort data= dataset;
by industry;
run;

proc sort data= dataset_1;
by industry;
run;

data dataset_2;
merge dataset dataset_1;
by industry;
run;

data dataset_3;
set dataset_2;
if n < 10 then delete;
run;
--------------------------------------------------------------

Re: Remove less than 10 observations

SPR — Thu, 03 Mar 2011 15:07:33 GMT

Hello q1234,

This is a possible solution:
[pre]
proc freq data=data noprint;
tables Industry*Year /out=f (where=(count > 10));
run;
proc SQL;
create table r as
select a.*
from data as a, f as b
where a.year=b.year and a.industry=b.industry
;quit;
[/pre]
Sincerely,
SPR

Re: Remove less than 10 observations

q1234 — Fri, 04 Mar 2011 11:17:45 GMT

Many thanks

Re: Remove less than 10 observations

FriedEgg — Fri, 04 Mar 2011 17:48:52 GMT

> data dataset_3;
> set dataset_2;
> if n < 10 then delete;
> run;

data dataset_3;
set dataset_2;
if _n_ < 10 then delete;
run;

Re: Remove less than 10 observations

Cynthia_sas — Fri, 04 Mar 2011 18:15:40 GMT

Hi:
An aside comment...if all you want to do is drop variables from the OUT= dataset, then this is OK syntax:
[pre]
proc tabulate data=dataset
out=dataset_1(drop=_table_ _type_ _page_);
class industry;
table industry*n;
run;
[/pre]

cynthia