topic Re: New Variable from Frequency Count in Statistical Procedures

New Variable from Frequency Count

buder — Tue, 10 Oct 2017 16:08:20 GMT

I am trying to create a new variable to be merged back into the old data set from a frequency count. Below is a snap-shot of the output. The new variable I want to create is a frequency count of DUPERSID.

DUPERSID	Frequency	Percent	Cumulative Frequency	Cumulative Percent
20002014	2	0	2	0
20002068	2	0	4	0
20003019	2	0	6	0.01
20003026	2	0	8	0.01
20003030	2	0	10	0.01
20004018	2	0	12	0.01
20007016	2	0	14	0.02
20007023	2	0	16	0.02
20007030	2	0	18	0.02
20012018	1	0	19	0.02
20012025	1	0	20	0.02
20013017	1	0	21	0.02
20014017	2	0	23	0.03
20014024	2	0	25	0.03
20015019	2	0	27	0.03

The code I used was as follows:

PROC FREQ DATA = TEST NOPRINT;

TABLES DUPERSID/

OUT=COUNTS (KEEP=DUPERSID COUNT RENAME=COUNT(DUPERSID_COUNTS));

RUN;

I did not receive an error code in log, however, when I run a proc freq:

PROC FREQ DATA = TEST;

TABLES DUPERSID_COUNTS;

RUN;

It says "variable not found." Any suggestions?

Re: New Variable from Frequency Count

Reeza — Tue, 10 Oct 2017 16:31:07 GMT

Your second PROC FREQ is still pointed at the TEST data set, but you called it COUNTS.

PROC FREQ DATA = TEST NOPRINT;

TABLES DUPERSID/

OUT=COUNTS (KEEP=DUPERSID COUNT RENAME=COUNT(DUPERSID_COUNTS));

RUN;

PROC FREQ DATA = TEST;

TABLES DUPERSID_COUNTS;

RUN;

An easier way to do this is to either merge it in or to use SQL.

proc sql;
create table want as
select *, count( dupersid) as count_dupersid
from have
group by dupersid;
quit;

Re: New Variable from Frequency Count

buder — Tue, 10 Oct 2017 17:02:56 GMT

Many thanks, got that to work out now. Did not know about PROC SQL, will try that as well.