topic Re: Compare columns and make one column in SAS Programming

Compare columns and make one column

weiwei2392 — Fri, 30 Dec 2016 08:56:42 GMT

Hi all the experts,

How do I find the earliest date within dateofadmin, dateofdrug and dateofmed for each ID using SAS code?

(ex. the earliest date for ID110 is dateofadmin14842 ((14842<18390 and ignore the ".")) )

and then i want to put the earliest in a new column :S

the date has been transfer into numeric format already.

However, i couldnt find the ways to compare numbers within columns :S

Please kindly give me a hint , would really be appreciated!

Re: Compare columns and make one column

Shmuel — Fri, 30 Dec 2016 09:29:34 GMT

data want;

set have;

firstdate = .;

array datex {3} dateofadmin dateofdrug dateofmed;

do i=1 to 3;

if datex(i) ne . then do;

if firstdate = . then firstdate = datex(i);

else if datex(i) < firstdate then firstdate = datex(i);

end; end;

run;

Re: Compare columns and make one column

Astounding — Fri, 30 Dec 2016 11:39:39 GMT

The MIN function should ignore missing values:

earliest_date = min(dateofadmin, dateofdrug, dateofmed);

Re: Compare columns and make one column

PGStats — Fri, 30 Dec 2016 20:39:58 GMT

The min function accepts any number of arguments.

data want;
set have;
firstdate = min(dateofadmin, dateofdrug, dateofmed);
run;

Re: Compare columns and make one column

weiwei2392 — Sun, 01 Jan 2017 05:25:51 GMT

thank you so much!

Now i have one more question

what if i wanna add one more column "where" is states where does the first date comes from?

like,

thank you 🙂

Re: Compare columns and make one column

Shmuel — Sun, 01 Jan 2017 05:50:55 GMT

data want;

set have;

firstdate = .;

array datex {3} dateofadmin dateofdrug dateofmed;

do i=1 to 3;

if datex(i) ne . then do;

if firstdate = . then do;

firstdate = datex(i);

where = vname(datex(i));

end;

else if datex(i) < firstdate then firstdate = datex(i);

end; end;

run;

Re: Compare columns and make one column

Astounding — Sun, 01 Jan 2017 13:59:53 GMT

It's possible to shorten the code by a line or two using arrays, but this variation might be easier to understand.

data want;

set have;

earliest_date = min(dateofadmin, dateofdrug, dateofmed);

length varname $ 32;

if earliest_date > . then select (whichn(earliest_date, dateofadmin, dateofdrug, dateofmed));

when (1) varname='dateofadmin';

when (2) varname='dateofdrug';

when (3) varname='dateofmed';

end;

run;