How do i identify first unique value by ID?

[lai302120]

Hi,

i'm trying to identify the distinct departments where revenue > 0 in the following dataset:

ID	Revenue	Dept	Flag (want to code)
1	$2	a	1
1	$1	a	0
1	$0	b	0
1	$3	c	1
1	$0	c	0
1	$5	d	1
1	$10	d	0
2	$2	a	1
2	$1	b	1
2	$0	c	0
2	$3	c	1
2	$0	d	0
3	$2	a	1
3	$1	a	0
3	$0	b	0
3	$3	a	0
3	$0	a	0

I only want to consider rows where REVENUE > 0. I also want to flag within each ID, the first unique department as FLAG = 1.

So, for ID = 1, the total number of unique departments is SUM(FLAG) = 3. 

Hope I made sense...please let me know if you need any additional info.

Thanks for your help!

[ballardw · Accepted solution]

If you data is sorted by the ID then there is a way to reference the first and last value of a group. The method requires using a BY statement;

For you problem does the existing order of the data matter or would it be okay to sort the data?

If sorting is okay;

proc sort data = have;
   by id dept descending revenue;
run;
 
data want;
   set have;
    by id dept;
    flag = first.dept and revenue>0;
run;

  will create a data set with the largest revenue flagged as 1 if greater than 0.

Your g

[lai302120]

Thanks this solution worked perfectly!
just a quick question..why do i need to sort revenue if there is already a "revenue>0" condition?

[ballardw]

The First. values are not conditional. By sorting Revenue descending then the largest Revenue value is associated with the First level of each Dept. Therefore if the first Revenue value is greater than 0 then you want the flag.

[lai302120]

Hi Ballard, 

i was wondering if i can get your help again..

i want my flag column to show the unique value instead of 1/0. is that easy to do? will the code be the same if i wanted numeric values? 

ID	Revenue	Dept	Flag (want to code)
1	$2	a	a
1	$1	a
1	$0	b
1	$3	c	c
1	$0	c
1	$5	d	d
1	$10	d
2	$2	a	a
2	$1	b	b
2	$0	c	c
2	$3	c
2	$0	d
3	$2	a	a
3	$1	a
3	$0	b
3	$3	a
3	$0	a