many to many merge using proc sql or datasteps?

[michtka]

Hi all,'ve got the next dataset (have):

data have;

length subjid 8 color1 color2 $20;

input subjid $ color1 $ color2;

datalines;

1 blue       blue

2 red       grey 

3 yellow   blue

4 green     green

5 black     yellow

6 white     pink

7 purple    yellow

;

run;

I am interested to create a final dataset (want) like this:

color2    other

blue     blue

grey    

green   green

yellow  yellow

pink  

as you can see I am interested to know which color in (color2) has been assigned to (color1), and which color in (color2) cant be assigned to (color 1) because  are different (blanck values)...in this example would be grey and pink.

Thanks in advance,

V.

[michtka]

maybe using arrays...please some help.

Thanks.

[Haikuo · Accepted solution]

How about this:

data have;

input (subjid color1 color2) (:$);

datalines;

1 blue blue

2 red grey 

3 yellow blue

4 green green

5 black yellow

6 white pink

7 purple yellow

;

proc sql;

  select distinct a.color2, b.color1 as other

  from have a

  left join

have b

on a.color2=b.color1

;

quit;

Haikuo

[michtka]

Brilliant Hai.Kuo

Many thanks.

Cheers,

V.