topic Re: Identifying earliest date across column indicators in SAS Programming

Identifying earliest date across column indicators

Walternate — Tue, 24 May 2016 15:43:58 GMT

Hi,

I have a dataset with ID and indicators for each month/year (=1 if person had an event in that month, otherwise =.).

ID event_200501 event_200502 event_200503.......event_200912

1 1 1 1 .

2 . . 1 1

I have created an array to identify the earliest month in which a person has an event in each year, like this:

data want;

set have;

array event2005(12) event_200501--event_200512;

do i=12 to 1;

if event2005(i)=1 then earliest_event_2005=i;

end;

/*Etc for all subsequent years*/

run;

The array runs with no error messages and earliest_event_2005 is created, but it's never actually populated (same with all other years). What I want is the exact same data but with new columns indicating the earliest event each person had in each year.

ID event_200501 event_200502 event_200503.......event_200912 earliest_event_2005....earliest_event_2009

1 1 1 1 . 1 .

2 . . 1 1 3 12

Any help is much appreciated.

Re: Identifying earliest date across column indicators

ballardw — Tue, 24 May 2016 16:05:46 GMT

Try this:

data want;
   set have;
   array  event2005(12) event_200501--event_200512;
   earliest_event_2005= whichn(min(of event2005(*)), of event2005(*));
run;

Re: Identifying earliest date across column indicators

FreelanceReinh — Tue, 24 May 2016 16:41:07 GMT

Hi @Walternate,

The issue with your code is that your iterative DO statement is incomplete:

do i=12 to 1 by -1;

Of course, ballardw's approach is more elegant.