How to calculate average mark for each student - postgresql

Question

I have a table "Register"
with columns:
class_id bigint NOT NULL,
disciple text,
datelesson date NOT NULL,
student_id integer NOT NULL,
note character varying(2)
now I want to calculate the average score for each student_id and the number of absent
Select * from "Register" as m
Join
(SELECT AVG(average), COUNT(abs) FROM (SELECT
CASE
WHEN "note" ~ '[0-9]' THEN CAST("note" AS decimal)
END AS average,
CASE
WHEN "note" ='a' THEN "note"
END AS abs
FROM "Register" ) AS average)n
on class_id=0001
And datelesson between '01.01.2012' And '06.06.2012'
And discipline='music' order by student_id
Result is this:
0001;"music";"2012-05-02";101;"6";7.6666666666666667;1
0001;"music";"2012-05-03";101;"a";7.6666666666666667;1
0001;"music";"2012-05-01";101;"10";7.6666666666666667;1
0001;"music";"2012-05-02";102;"7";7.6666666666666667;1
0001;"music";"2012-05-03";102;"";7.6666666666666667;1
0001;"music";"2012-05-01";102;"";7.6666666666666667;1
The result I receive is for the whole column but how do I calculate average marks for each student?

Answer 1

Could look like this:
SELECT student_id
, AVG(CASE WHEN note ~ '^[0-9]*$' THEN note::numeric
ELSE NULL END) AS average
, COUNT(CASE WHEN note = 'a' THEN 1 ELSE NULL END) AS absent
FROM "Register"
WHERE class_id = 1
AND datelesson BETWEEN '2012-01-01' AND '2012-06-06'
AND discipline = 'music'
GROUP BY student_id
ORDER BY student_id;
I added a couple of improvements.
You don't need to double-quote lower case identifiers.
If you want to make sure, there are only digits in note, your regular expression must be something like note ~ '^[0-9]*$'. What you have only checks if there is any digit in the string.
It's best to use the ISO format for dates, which works the same with any locale: YYYY-MM-DD.
The count for absence works, because NULL values do not count. Ypu could also use sum for that.
As class_id is a number type, bigint to be precise, leading zeros are just noise.
Use class_id = 1 instead of class_id = 0001.

Answer 2

Looks to me like you are missing a "group by" clause. I am not familiar with postgress but I suspect the idea applies just the same.
here is an example in transact-sql:
--create table register
--(
--class_id bigint NOT NULL,
-- disciple text,
-- datelesson date NOT NULL,
-- student_id integer NOT NULL,
-- grade_report integer not null,
-- )
--drop table register
delete from register
go
insert into register
values( 1, 'math', '1/1/2011', 1, 1)
insert into register
values( 1, 'reading', '1/1/2011', 1, 2)
insert into register
values( 1, 'writing', '1/1/2011', 1, 5)
insert into register
values( 1, 'math', '1/1/2011', 2, 8)
insert into register
values( 1, 'reading', '1/1/2011', 2, 9)
SELECT student_id, AVG(grade_report) as 'Average', COUNT(*) as 'NumClasses'
from register
WHERE class_id=1
group by student_id
order by student_id
cheers