Count by Variable Groups

[Barkamih]

Hi all 

I would like to know if this code is correct for my question.

I want to know the number of herds in each month for each county by using this code.

regards 

Ibrahim 

Proc freq data = have;
tables month*county*herd / out = want;
run; quit; 

[RW9]

So long as the by groups are in order (I would guess):

proc freq data=have;
  tables county*month*herd / out=want;
run;

 Note you don't need the quit;

[Astounding]

No, not quite.  It tells you number of times each herd appeared in each month/county.  You would have to continue with:

proc freq data=want;

tables month*county / out=truly_want;

run;

That summarizes to the counts that you are looking for:  number of herds for each month/county.

[Reeza]

It depends on your data structure and data. If you have repeated measurements the answer is different than if you have a single list of entries. 

[Barkamih]

Sure, that is my issue. For example, the county is repeated many times by year, as well as herds are repeated by year for each county. So in this case what is the code looks like?

regards 

[Reeza]

Post sample data please, fake data is fine.

---

@Barkamih wrote:

Sure, that is my issue. For example, the county is repeated many times by year, as well as herds are repeated by year for each county. So in this case what is the code looks like?

regards 

---

[Barkamih]

herd	     County	month	YEAR
59994724	Lim	3	2004
59994724	Lim	7	2004
59994724	Lim	4	2004
59994724	Lim	8	2004
59994724	Lim	9	2004
59994724	Lim	5	2004
59994724	Lim	6	2004
59994724	Lim	7	2005
59994724	Lim	10	2005
59994724	Lim	9	2005
59994724	Lim	12	2005
59994724	Lim	6	2005
59994724	Lim	8	2005
59994724	Lim	11	2005
59994724	Lim	8	2004
59994724	Lim	7	2004
59994724	Lim	6	2004
59994724	Lim	2	2004
59994724	Lim	5	2004
59994724	Lim	4	2004
59994724	Lim	2	2005
59994724	Lim	3	2005
59994724	Lim	5	2005
59994724	Lim	10	2005
59994724	Lim	9	2005
59994724	Lim	8	2005
59994724	Lim	7	2005
59994724	Lim	6	2005
60005584	Co	4	2004
60005584	Co	5	2004
60005584	Co	6	2004
60005584	Co	10	2004
60005584	Co	8	2004
60005584	Co	9	2004
60006084	Co	10	2004
60006084	Co	7	2004
60006084	Co	11	2004
60006084	Co	4	2004
60006084	Co	9	2004

[Reeza · Accepted solution]

proc sql;
create table want as
select county, month, year, count(distinct herd) as n_herd
from have
group by county, month, year;
quit;

---

@Barkamih wrote:

herd	     County	month	YEAR
59994724	Lim	3	2004
59994724	Lim	7	2004
59994724	Lim	4	2004
59994724	Lim	8	2004
59994724	Lim	9	2004
59994724	Lim	5	2004
59994724	Lim	6	2004
59994724	Lim	7	2005
59994724	Lim	10	2005
59994724	Lim	9	2005
59994724	Lim	12	2005
59994724	Lim	6	2005
59994724	Lim	8	2005
59994724	Lim	11	2005
59994724	Lim	8	2004
59994724	Lim	7	2004
59994724	Lim	6	2004
59994724	Lim	2	2004
59994724	Lim	5	2004
59994724	Lim	4	2004
59994724	Lim	2	2005
59994724	Lim	3	2005
59994724	Lim	5	2005
59994724	Lim	10	2005
59994724	Lim	9	2005
59994724	Lim	8	2005
59994724	Lim	7	2005
59994724	Lim	6	2005
60005584	Co	4	2004
60005584	Co	5	2004
60005584	Co	6	2004
60005584	Co	10	2004
60005584	Co	8	2004
60005584	Co	9	2004
60006084	Co	10	2004
60006084	Co	7	2004
60006084	Co	11	2004
60006084	Co	4	2004
60006084	Co	9	2004

---

[Barkamih]

thanks very much my friend
I appreciate it
best regards

Ibrahim