Cleaning up Arrays

[ritam2011]

Guys, I have an issue where I have to delete duplicate elements of a character array.

For example:

Array1:  [cat dog hen cat dog] ----> [cat dog hen]

or

Array2:  [cat cat cat] ----> [cat]

I want to maintain the original order.  Any ideas?

[Reeza]

What about all the ladies? :smileyangry:

Use a loop that searches the rest of the array for the same value, if the value is the same set it to missing instead. 

Or use whichc to find the other variables and the index for it to again set it to missing. 

In the end you may need to shorten your array by dropping some variables.

There was a solution posted recently for a similar problem but with airplane codes and looking for flight paths I think....

[data_null__]

WhichC sounds like a good idea.

323  data test;

324     array c[5] $3;

325     input c

;

326     put;

327     put 'NOTE: Before (' (c

)(+1 $3.) ' )';

328     do i = 1 to dim(c);

329        if i gt whichC(c,of c

) then call missing(c);

330        end;

331     put 'NOTE- After  (' (c

)(+1 $3.) ' )';

332     cards;

NOTE: Before ( cat dog hen cat dog )

      After  ( cat dog hen         )

NOTE: Before ( cat cat cat         )

      After  ( cat                 )

[DLing · Accepted solution]

Here's a really dumb idea but it should work:

do i = 1 to N;  (for each element of array - N is number of elements of array)

     if carray ~= '' then do;

          do j = i+1 to N;     (for all following elements)

               if carray = carray then carray = '';   (wipe out all elements same as carray)

          end;

     end;

end;

At the end of this loop, you have an array with holes in it, but dup elements are gone.

If you want, you can squish the array so that the elements are contiguous via:

pos = 1;

do i = 1 to N;

     if carray ~= '' then do;

          carray[pos] = carray;

          pos = pos + 1;

     end;

end;

Not elegant, but should work.

Another idea is, if instead of array, the source is a really long string, then tranwrd() can be used to eliminate words without looping.  This makes the dup word elimination easier, but you pick up other headaches of keeping track of what's been done already,... etc.  Don't think this will be more efficient to execute.

[Ksharp]

Why not use Hash Table which has this ability.

Ksharp

[ritam2011]

Thanks everyone for your help.  I ended up using D-Ling's suggestions for the nested DO loop to check all values on each row and replacing duplicates with a missing value. One thing I did notice (prob since I'm new to this) was that I had to use ' ' instead of . to replace a value with a missing value or else the column width became extremely large.