Collate variable names to new variable if hit

[danhopkinslewis]

Hi,

Wonder if you can help. I'm trying to create a new variable which brings in variable names where they equal 1.

For example Reason is the variable I want to create.

Application_number	A	B	C	Reason
5002	1	1	0	A,B
5003	1	1	1	A,B,C
5004	0	0	1	C
5005	1	1	1	A,B,C

I've tried a VNAME but could only get it to bring in 1 hit. My data source has 49 variables that could have a possible 1.

Thanks

Dan

[PeterClemmensen · Accepted solution]

Try this

data have;
input Application_number A B C;
datalines;
5002 1 1 0 
5003 1 1 1 
5004 0 0 1 
5005 1 1 1 
;

data want;
   set have;
   array rr A B C;
   length Reason $ 100;
   do over rr;
      if rr then Reason = catx(',', Reason, vname(rr));
   end;
run;

Result:

Application_number  A  B  C  Reason 
5002                1  1  0  A,B 
5003                1  1  1  A,B,C 
5004                0  0  1  C 
5005                1  1  1  A,B,C 

[danhopkinslewis]

Works a treat  @PeterClemmensen massive appreciation sent your way.

[PeterClemmensen]

Massive appreciation appreciated 🙂

[PaigeMiller]

---

@danhopkinslewis wrote:

Hi,

Wonder if you can help. I'm trying to create a new variable which brings in variable names where they equal 1.

For example Reason is the variable I want to create.

Application_number	A	B	C	Reason
5002	1	1	0	A,B
5003	1	1	1	A,B,C
5004	0	0	1	C
5005	1	1	1	A,B,C

I've tried a VNAME but could only get it to bring in 1 hit. My data source has 49 variables that could have a possible 1.

---

I curious, what could you possibly do with a string such as one you are trying to create, if it has 49 different possible variables? What benefits would there be, compared to working with the 0/1 variables?