|
On Tue, 3 Oct 2006 10:01:08 -0400, Peter Crawford
<peter.crawford@BLUEYONDER.CO.UK> wrote:
>On Tue, 3 Oct 2006 06:53:31 -0700, Newbie <oldscot82@YAHOO.COM> wrote:
>
>>data test;
>>input id$ code$;
>>datalines;
>>1111 apple
>>1111 orange
>>1111 lemon
>>2222 green
>>2222 yellow
>>2222 banana
>>2222 pink
>>3333 apple
>>3333 yellow
>>3333 orange
>>4444 grey
>>4444 white
>>4444 orange
>>5555 red
>>5555 blue
>>5555 green
>>;
>>run;
>>/*
>>Hi,
>>In the above dataset, I am trying to extract all ID's that
>>have any of these codes(apple,orange,lemon).
>>If the ID has any of those codes, I would like to extract all records
>>for that ID;
>>Thanks for your help.
>>
>>My final dataset would look like this.
>>1111 apple
>>1111 orange
>>1111 lemon
>>3333 apple
>>3333 yellow
>>3333 orange
>>4444 grey
>>4444 white
>>4444 orange
>>
>>Here is what I have got so far
>>
>>proc sort data=test;
>>by id;
>>run;
>>
>>data test2 ;
>>set test;
>>by id;
>>retain count;
>>if first.id then do;
>>count=0;
>>end;
>>if code in ('apple','orange','lemon') then do ;
>>count+ 1;
>>end;
>>
>>run;
>
>time to learn about
> where
>
>data wanted ;
> set test ;
> where code in ('apple','orange','lemon') ;
>run;
>
>Of course, if instead of the original data, you want counts,
>put that "where statement" into proc freq
>
>Good Luck
>
>Peter Crawford
sorry...
mea culpa
this old scot still needs to learn to read before responding !
Not just individuals where found, but the whole ID-group when
any in the group are found .
Multi-pass is perhaps neccessary, but since the data is in ID-order
this extract can merge fruit wanted, with all the details, by ID
data test2 ;
merge test( in= wanted
where= ( code in ('apple','orange','lemon') )
)
test ;
by ID ;
if wanted ;
run;
Peter
|
