05-19-2013 11:38 PM
I have a variable delivery_date with one observation and test_date with 20 observations. I'd like to find the closest date in test_date to delivery_date by find the minimum absolute difference and retain the minimum value.
When I merge the datasets I only get two observations paired up and get missing values for the rest. Would anyone please explain how to go about this with maybe a do loop or...???
05-20-2013 01:09 AM
You cannot merge by date as the dates do not match.
You could create a dummy variable that is a constant in each and merge on that (not really needed, but it you had a grouping variable it might help).
You could just set the single obs dataset once.
if _n_=1 then set delivery_dataset;
diff = delivery_date - test_date;
absdiff = abs(diff);
Probably easiest to just use PROC SQL.
proc sql noprint ;
create table want as
select a.delivery_date,b.test_date,a.delivery_date-b.test_date as diff, abs(a.delivery_date - b.test_date) as absdiff
from delivery_dataset a , test_dataset b
order by 4