Array

[TH2000]

I have a data set similar to the example at the bottom.  The example represent two sets of variables.  Set #1 is advfnl1 - advfnl6 and set #2 is advfnl1_c - advfnl6_c.

My goal is to create 4 new variables and take the top 2 from the advfnl1 - advfnl6 set and the top 2 from the advfnl1_c - advfnl6_c set...But the top two from each set cannot be a duplicate in the final 4 values saved.

The hierachy is as follows:

1) Always take the first value from advfnl1 and place into "NEWVAR1".

2) The next value comes from advfnl1_c you would place that in 'NEWVAR2' unless it's the same value as 'NEWVAR1'.  If it would be the same value as 'NEWVAR1' then use the value that is in advnl2_c and place that value in "NEWVAR2".

3) Place the value of advfnl2 in "NEWVAR3" unless it's the same value as 'NEWVAR1' or 'NEWVAR2'.  If it is the same value then use advfnl3.

4) Place the value of advfnl2_c in "NEWVAR4" unless it's the same value as 'NEWVAR1' or 'NEWVAR2' or 'NEWVAR3'.  If it is the same value then ues advfnl3_c.

Using the example below (and my logic above) the final results would look like:

NEWVAR1	NEWVAR2	NEWVAR3	NEWVAR4
34	40	29	49

ID	advfnl1	advfnl2	advfnl3	advfnl4	advfnl5	advfnl6	advfnl1_c	advfnl2_c	advfnl3_c	advfnl4_c	advfnl5_c	advfnl6_c
1234	34	29	43	38	12	54	40	29	34	49	54	22

Let me know if you can assist me in building this array.

Thanks.

[data_null__]

Based on what I think you want this is what I came up with.

data id;
   length id $4;
   array new[4];
   array advfnl[6];
   array c_advfnl[6];
   input ID advfnl

c_advfnl

;

   j = 0;
   do i = 1 to dim(advfnl);
      y = whichn(advfnl,of new

);
     

if not y then do; j+1; new=advfnl; end;
      if j eq dim(new) then leave;

      do k = 1 to dim(c_advfnl);    
         x=whichn(c_advfnl,of new

);

         if not x then do; j+1; new=c_advfnl; leave; end;
         end;
      if j eq dim(new) then leave;
      end;     
   drop i j k x y;
   cards;
1234 34 29 43 38 12 54   40 29 34 49 54 22
1235 34 29 43 38 12 54   34 29 34 49 54 22
1236 34 29 43 38 12 54   34 29 34 29 34 22
1237 34 29 34 29 12 54   34 29 34 29 34 22
1238 34 29 34 29 12 54   34 34 34 34 34 34
1239 34 29 34 29 29 29   34 34 34 34 34 34
;;;;
   run;
proc print heading=v;
   run;

[Fugue]

Your text explanation does not match your expected results:

You state newvar 4 should be 49.

However, your explanation for newvar 4 is: "Place the value of advfnl2_c in "NEWVAR4" unless it's the same value as 'NEWVAR1' or 'NEWVAR2' or 'NEWVAR3'.  If it is the same value then ues advfnl3_c.".

In your example, advfnl2_c = 29 and that matches newvar3. According to your instructions, we should, therefore, set newvar4 to advfnl3_c (=34).

Either your expected results are wrong, or your explanation is wrong.

Regardless, you could something like the following logic (which adheres to your explanation -- not your expected result). I've re-named your variables for convenience:

data new ;

merge old1 old2;

by idno;

format newvar1-newvar4 3.;

newvar1 = advfnl_1;

if advfnl_c1 = newvar1 then newvar2 = advfnl_c2;

else newvar2 = advfnl_c1 ;

if advfnl_2 = newvar1 or advfnl_2 = newvar2 then newvar3 = advfnl_3;

else newvar3 = advfnl_2;

if advfnl_c2 = newvar1 or advfnl_c2 = newvar2 or advfnl_c2 = newvar3 then newvar4 = advfnl_c3;

else newvar4 = advfnl_c2;

run;