01-23-2017 12:28 PM - edited 01-23-2017 12:41 PM
I have a table with a dummy var set to 0 for all rows EXCEPT the rows most recently sent for each person. In that case the dummy var should be set to 1.
The dummy var is incorrectly set to 1 for more than just the most recently sent row so I need to reset the dummy var to zero for all rows EXCEPT the rows most recently sent by each person.
I have a var called report_time and can ID the most recently sent row by using max(report_time).
The following syntax did not work:
set dummy_var =
when max(report_time) = report_time then dummy_var=1
else dummy_var = 0
group by person
order by report_time desc, person asc;
below end is the following:
ERROR: The value expression referenced by column dummy_var can not directly contain a summary function.
Below group is the following:
ERROR 180-322: Statement is not valid or it is used out of proper order.
01-23-2017 01:01 PM
You can do it easyly by sort and a data step:
proc sort data=have; by person_id report_time; run;
if last.person_id then dummy_var=1;
01-23-2017 02:42 PM
01-30-2017 09:39 AM
I tested the timing of doing a proc sql update of the rows that needed to be changed from dummy_var = 1 to dummy_var = 0, and found that takes over an hour which is much too long.
The next approach I tested was to delete the rows that needed updating, then adding those rows back in but with the correction.
I was surprised to find it takes 10 times longer to delete rows using proc sql than it takes using a data step.
Using proc sql insert is very quick so that is the method I used to get the rows of data back into the dataset.
I am trying to update an oracle table using base SAS in the most efficient manner and welcome any suggestions.
02-01-2017 04:31 AM