Replacing null value in SAS with next available value by group - arrays

Question

I am trying to replace missing values that occur before the first non-null entry in SAS. I have the following data:
StudentID Day TestScore
Student001 0 .
Student001 1 78
Student001 2 89
Student002 3 .
Student002 4 .
Student002 5 .
Student002 6 95
I'd like to modify the data so the null values are replaces with the next available non-null entry:
StudentID Day TestScore
Student001 0 78
Student001 1 78
Student001 2 89
Student002 3 95
Student002 4 95
Student002 5 95
Student002 6 95

Answer 1

data scores;
length StudentID $ 10;
input StudentID $ Day TestScore;
datalines;
Student001 0 .
Student001 1 78
Student001 2 89
Student002 3 .
Student002 4 .
Student002 5 .
Student002 6 95
;
run;
proc sort data = scores;
by descending day;
run;
data scores;
drop addscore;
retain addscore;
set scores;
if testscore ne . then addscore = testscore;
if testscore eq . then testscore = addscore;
run;
proc sort data = scores;
by day;
run;

Answer 2

proc sort data = have;
by id descending day ;
run;
data want;
set have;
by id;
retain last_score;
if first.id then call missing(last_score);
if not missing(score) then last_score = score;
else score = last_score;
run;
proc sort data=want;
by id day;
run;
FYI, this will NOT set the missing values if there are any after the last known score for a given ID. i.e. if you had something like:
Student002 5 95
Student002 6 .
Then only records prior to day 5 for id 002 will get a value of 95. Is that a possible condition for you? If yes, this solution will require a slight modification

Answer 3

You can use a DOW loop to identify the next non-missing score, and a subsequent DOW loop to apply the non-missing score. The DOW approach does not require sorting and maintains the original row order.
data want;
do _n_ = 1 by 1 until (last.id or not missing(score));
set have;
by id;
end;
_score = score;
do _n_ = 1 to _n_;
set have;
score = _score;
output;
end;
drop _score;
run;
In SQL, presuming day ordering, the imputed value can be looked up in a correlated sub-query.
proc sql;
create table want as
select
id, day,
case
when not missing(score) then score
else (select score from have as inner
where inner.id = outer.id
and inner.day > outer.day
and not missing(score)
having inner.day = min(inner.day)
)
end as score
from have as outer;