PROC SQL - Concatenate/transpose several numeric columns

[Adir92]

Hello everyone, 

I have a table like this : 

//	//	App1	App2	Pj1	Pj2	//
//	//	2	6	.	.	//
//	//	.	.	5	9	//
//	//	.	.	4	1	//

And I want to have this : 

//	//	AP	AP2	//
//	//	2	6	//
//	//	5	9	//
//	//	4	1	//

The idea is to concatenate the columns. 

Do I have to use cat functions here ? Because I have several columns to use and I don't know how to apply this in my case. 

// = random numbers, not important 

[ballardw]

It might help to describe which "column" is concatenated with which.

It might be that you are looking for COALESCE but with out stating which "columns" are combined to get the output I can't be sure. Example might be

                    SQL: Coalesce (AP, Pji) as Ap.

              data step:   Ap = Coalesce (AP, Pji);

Coalesce goes through a list of variables from left to right and returns the first non-missing(if any) value.

And

Not a transpose and SQL is generally a poor tool for transpose.

[Adir92]

Thank you,

I want to App1 and Pj1 ---> AP ; App2 and Pj2 ----> AP2

[andreas_lds · Accepted solution]

And what should happen if more than one value is non-missing?

In the meantime try

data want;
  set have;
  /* assuming all vars are numeric */
  ap1 = coalesce(App1, Pj1);
  ap2 = coalesce(App2, Pj2);

  /* maybe drop the old vars */
  drop App1 App2, Pj1 Pj2;
run;

[Adir92]

There's always a value missing in a row. 

Thank you very much ! I think this will work.