how to create dummy observations for each subject

[HitmonTran]

hello,

I want to output observation "Visit" for all subjects.  

data have:

subject	visit
001	monday
001	friday
002	monday
003	monday
003	tuesday
003	wednesday

data want:

subject	visit
001	monday
001	tuesday
001	wednesday
001	thursday
001	friday
002	monday
002	tuesday
002	wednesday
002	thursday
002	friday
003	monday
003	tuesday
003	wednesday
003	thursday
003	friday

[Tom · Accepted solution]

You need a list of subjects and a list of visits.  

Let's assume the list of subjects comes from the dataset you already have.

What is the source for the list of visits?

Perhaps something like this:

data visits;
  input visit $20.;
cards;
monday
tuesday
wednesday
thursday
friday
;

So a simple join like this should generate what you want.

proc sql;
create table want as select *
  from (select distinct subject from have)
     , vists
  order by subject, visit
;
quit;

[mkeintz]

You apparently want a Cartesian crossing of subject and visit (although you don't have Thursday in your sample data, but do have it in your desired output.  Use the "completetypes" option of PROC SUMMARY:

data have;
  input subject visit :$9.    ; 
  format subject z3.;
datalines;
1 Monday
1 Friday
2 Monday
3 Monday
3 Tuesday
3 Wednesday
run;

proc summary data=have completetypes nway;
  class subject visit ;
  output out=want (drop=_type_);
run;