keeping observation based on two variables

[lillymaginta1]

data have;
infile cards ;
input ID    matchto  kp ;	 
cards;
001 1 1 
002 2 0
003 3 1 
004 1 0
005 2 0
006 3 0
;

I have the data above. I am trying to create a subset based on the following two conditions: a) either kp=1 then keep or b) if kp=0 but the matchto variable equal to another one where kp=1 then also keep. For example, keep id 004 and 006 since the value of "matchto" equal to the one where kp=1.

The output  data

001 1 1 003 3 1 004 1 0 006 3 0 

[lillymaginta1]

Sorry, the output data is the following 

ID    matchto  kp 	 
001 1 1 
003 3 1 
004 1 0
006 3 0

[Reeza · Accepted solution]

SQL is great here.

proc sql;
select * from have
where kp=1 or matchto in (select distinct matchto from have where kp=1);
quit

---

@lillymaginta1 wrote:

data have;
infile cards ;
input ID    matchto  kp ;	 
cards;
001 1 1 
002 2 0
003 3 1 
004 1 0
005 2 0
006 3 0
;

I have the data above. I am trying to create a subset based on the following two conditions: a) either kp=1 then keep or b) if kp=0 but the matchto variable equal to another one where kp=1 then also keep. For example, keep id 004 and 006 since the value of "matchto" equal to the one where kp=1.

The output  data

001 1 1 003 3 1 004 1 0 006 3 0 

---

[novinosrin]

Elegant and super fast solution @Reeza 

Just adding variety to menu-

data have;
infile cards ;
input ID    matchto  kp ;	 
cards;
001 1 1 
002 2 0
003 3 1 
004 1 0
005 2 0
006 3 0
;

data want;
 if _n_=1 then do;
  dcl hash h(dataset:'have(where=(kp))');
  h.definekey('matchto');
  h.definedone();
 end;
 set have;
 if kp or h.check()=0;
run;

[lillymaginta1]

Thank you both!