Find all possible groups of pairs of two variables

[PhilC]

Wow,

~100 variables.  Cases like these are when I've found its nice to learn computer science.  You'll never forget that first time you unknowingly start down a rabbit hole that needs O(n!) time.  Asymptotic computational complexity - Wikipedia

[Joe_O]

I do mean observations but once they are read into the array they will become variables as input to the function.

[Tom]

---

@Joe_O wrote:
I do mean observations but once they are read into the array they will become variables as input to the function.

---

https://xyproblem.info/

Perhaps you should explain the real problem you are trying to solve.  There might already be a SAS procedure that has been optimized for this.

[Joe_O]

Perhaps you're right. Would you advise creating a new topic or updating / replying within this one?

[PaigeMiller]

---

@Joe_O wrote:
I do mean observations but once they are read into the array they will become variables as input to the function.

---

The above statement (and really all your statements) still does not have enough meaning to turn into a program, in my opinion. I still look forward to seeing a clear and complete problem description, with sample input data (in the format requested earlier) and step-by-step description of how the variable COMBINATION is created.

[PhilC]

I liked this problem.  I learned about a new function: lexperm  I hope this adds to the discussion, even though you've found a solution,

data temp;
   array y_[3] $1 y1-y3 ('1' '2' '3' );
   nfact=fact(3); drop nfact;
   do i=1 to nfact;
      rc=lexperm(i, of y_[*]);
        drop  rc;
      if rc<0 then leave;
     output;
   end;
run;
data want (Keep=Combination x y);
  set temp;
    array y_[3] y1-y3;
  Combination="Combination"||put(i,1.0);
  do x=1 to 3;
    y=y_[x];
    output;
  end;
run;

[Joe_O]

Decided to mark this as the solution since it is exactly what I was looking for and scaling it up was never realistic! Thanks for your help. This will definitely be useful in future.