differen value within second by variable

[SASSLICK001]

Hi I have the following 

id1 id2

01  1

01  1

01  2

02  1

02  1

02  1

How can I ideintify the different value id2 within id1?

[Astounding]

This both identifies and counts for you:

proc freq data=have;

tables id1 * id2 / list missing;

run;

[SASSLICK001]

I wanted to use first. and last.

[ballardw]

---

@SASSLICK001 wrote:
I wanted to use first. and last.

---

It would help to show what you expect for output.

Maybe something like ?

data want;
   set have;
   by id1 id2;
   if first.id2;
run;

[SASSLICK001]

I want to create a flag where the value changes per change in value for id2

 expected output

id1 id2  flag

01  x      1

01  x      1 

01  y      2

02  z      1

02  z      1

02  z      1

[ballardw]

Assuming your data is sorted by ID1 and ID2 as implied by your example:

data have;
input id1 $ id2  $;
datalines;
01  x 
01  x 
01  y 
02  z 
02  z 
02  z 
;
run;
data want;
   set have;
   by id1 id2;
   if first.id1 then flag=0;
   else flag = first.id2;
run;

I prefer the 0/1 coding as 0 means 'not a change' in this case and there are some advantages that I can count the number of changes in id1 by getting a sum of the flag variable.

If your data is not sorted then you need to provide a more realistic example of the data.

[Astounding · Accepted solution]

For that result:

proc sort data=have;

by id1 id2;

run;

data want;

set have;

by id1 id2;

if first.id1 then flag=1;

else if first.id2 then flag + 1;

run;