topic Re: Count based on groupings by unique IDs in SAS Programming

Count based on groupings by unique IDs

tm28 — Tue, 09 Apr 2019 14:12:44 GMT

I'm having difficulty with a pretty easy procedure. I'm trying to get the number of records based on groups (O, IV and Combination of O+IV).

I have the following data.

data have;
infile datalines delimiter=",";
input ID, recordkey, route $;
datalines;
1,1,IV
1,2,IV
1,3,IV
2,4,O
2,5,IV
2,6,IV
3,7,O
3,8,O
;

How would I get a count of unique IDs that are O alone, IV alone and combination of O and IV?

I want to get the following results:

Oral=1

IV=1

Combo=1

Re: Count based on groupings by unique IDs

novinosrin — Tue, 09 Apr 2019 13:55:13 GMT

do these

3,7,O
3,8,0

belong to O?

And count of distinct ids for IV seems 2 , are you sure it's IV=3????

Re: Count based on groupings by unique IDs

tm28 — Tue, 09 Apr 2019 14:00:19 GMT

Sorry, I updated the error in the initial code on the last line.

Re: Count based on groupings by unique IDs

novinosrin — Tue, 09 Apr 2019 14:24:14 GMT

Hi @tm28 a very interesting problem. Certainly not that simple for me as it is probably for you

data have;
infile datalines delimiter=",";
input ID recordkey route $;
datalines;
1,1,IV
1,2,IV
1,3,IV
2,4,O
2,5,IV
2,6,IV
3,7,O
3,8,O
;

proc sql;
create table want as
select grp, count(distinct id) as count
from
(select *,ifc(count(distinct route) =1 ,route,'combo') as grp from have group id)
group by grp;
quit;

Re: Count based on groupings by unique IDs

tm28 — Tue, 09 Apr 2019 14:31:05 GMT

This works great for the most part, except for O route.. It should have a count of 1, not 2.

Re: Count based on groupings by unique IDs

novinosrin — Tue, 09 Apr 2019 14:32:40 GMT

@tm28 Here is my test:

data have;
infile datalines delimiter=",";
input ID recordkey route $;
datalines;
1,1,IV
1,2,IV
1,3,IV
2,4,O
2,5,IV
2,6,IV
3,7,O
3,8,O
;

proc sql;
create table want as
select grp, count(distinct id) as count
from
(select *,ifc(count(distinct route) =1 ,route,'combo') as grp from have group id)
group by grp;
quit;

proc print noobs;run;

grp	count
IV	1
O	1
combo	1

Re: Count based on groupings by unique IDs

Ksharp — Tue, 09 Apr 2019 15:06:57 GMT

data have;
infile datalines delimiter=",";
input ID recordkey route $;
datalines;
1,1,IV
1,2,IV
1,3,IV
2,4,O
2,5,IV
2,6,IV
3,7,O
3,8,O
;

proc sql;
create table want as
select 'Oral' as name,count(distinct id) as n from have group by id having sum(route='O')=count(*)
union
select 'IV' as name,count(distinct id) from have group by id having sum(route='IV')=count(*)
union
select 'Combo' as name,count(distinct id) from have group by id having count(distinct route)>1;
quit;

Re: Count based on groupings by unique IDs

tm28 — Tue, 09 Apr 2019 17:16:38 GMT

This got to what I wanted (minus the union part of the code), thanks so much!