Subset data with number of records count

[zimcom]

data want; set have;
retain z;
if first.SUBJECT then z = 1;
else z = z + 1;
by SUBJECT;
run;

After I run the above code, how can I remove the SUBJECT if last.SUBJECT is less than 3 (not the observation of "z LT 3", but the SUBJECT), which means if the SUBJECT has less than 4 records, then they will be deleted.

Thanks as always

[Kurt_Bremser · Accepted solution]

proc sql;
create table want as
  select *
  from have
  where subject not in (
    select subject from have group by subject having count(subject) le 4
  )
;
quit;

[zimcom]

Thank you so much, really appreciated!

[zimcom]

Another related question, in the dataset, now each subject has at least 4 records (up to the maximum of 6). I want to do another subset of only those subjects where the difference between the 2nd to 3rd, 3rd to 4th, 4th to 5th, etc., jumped great than 4 points from the previous one. 

Thanks again!