Merging Issue

[gamotte]

As you can see, the array is declared as
array a_mngrs {*} a--c;
That is all columns from a to c. If you have more managers, just indicate the first and the last instead of a and c.

[wizkid2050]

That's fine but what if dont want to keep changing the first and last values. may be i just want my code to work for all datasets thrown at it.

[gamotte]

You can use _NUMERIC_ instead of a--c to select all numeric columns of the transposed dataset.

[wizkid2050]

Awesome work man.!! Kudos. It works.

[wizkid2050]

But i tried adding d with count 42 to check. its only giving 10 observations in output.
Just check with this condition.

[gamotte · Accepted solution]

data test3;
	set tr_test2;
	/*keep managers;*/
	array a_mngrs {*} _NUMERIC_;

	if _N_=1 then index=1;

	do i=1 to &n_accounts.;
		if min(of a_mngrs{*}) lt 50 then do;
			do j=1 to dim(a_mngrs);
				if a_mngrs{index} lt 50 then do;
					a_mngrs{index}=a_mngrs{index}+1;
					managers=vname(a_mngrs{index});
					j=dim(a_mngrs); /* Exit loop on j*/
					output;
				end;

				/* Probably a better way to do this */
				index=mod(index+1,dim(a_mngrs));
				if index=0 then index=dim(a_mngrs);
			end;
		end;
		else do;
			i=&naccounts.; /* Exit loop on i */
		end;
	end;
run;

[wizkid2050]

Now this is exactly i was looking for. Great job.

[gamotte]

You're welcome. There is still a typo you probably noticed

		else do;
			i=&n_accounts.; /* Exit loop on i */
		end;

I forgot the underscore in &n_accounts.

Also, you can uncomment the keep that i commented for debug.

[Astounding]

Explain the stopping condition a bit more ... how does count reach 50?

If a solution involved sorting (and yet still got the answer you wanted) is that feasible or is your data set too large?