Get the sum of all rows only once per userid - mysql

Question

I am getting the sum of all the rows in registration table. Table structure looks like this:
row_id row_registration_fee row_amount row_class row_user_id
1 200 1000 18 1
2 200 2510 18 1
3 0 1600 19 2
4 0 1500 19 1
5 200 1254 19 3
6 200 3000 19 1
7 200 2000 19 1
8 0 100 20 1
9 0 300 20 2
A user can have multiple registration fee. And I need to get the sum of all the row_registration_fee by row_class. The result should be like this:
row_registration_fee row_class
200 18
400 19
0 20
My select :
SELECT (COUNT(DISTINCT(row_user_id))* 200) as 'fee'
FROM registration
WHERE row_registration_fee > 0
GROUP BY row_class
Is there a better query here that can give a result like the above sample?
The result will be displayed in table rows using foreach loop in PHP. As of now, it will only give me two results for the registration fee, the row_class 18 and row_class 19 it excludes the row_class 20 because it only selects the user with a fee.
Additional explanation: The user with 2 or more registration fees should count only as one if the user has a total of 400 fees it should sum only 200.

Answer 1

Another approach might be to find the most recent row_id. In this case I have altered the data so that the first entry appears to be an error (fee = 300) followed by the second entry for the correct amount .
drop table if exists t;
create table t(
row_id int,row_registration_fee int,row_amount int, row_class int, row_user_id int);
insert into t values
(1 , 300 , 1000 , 18 , 1),
(2 , 200 , 2510 , 18 , 1),
(3 , 0 , 1600 , 19 , 2),
(4 , 0 , 1500 , 19 , 1),
(5 , 200 , 1254 , 19 , 3),
(6 , 200 , 3000 , 19 , 1),
(7 , 200 , 2000 , 19 , 1),
(8 , 0 , 100 , 20 , 1),
(9 , 0 , 300 , 20 , 2)
;
select sum(row_registration_fee),row_class
from
(
select t.row_class,t.row_registration_fee
from t
where t.row_id = (select max(row_id) from t t1 where t1.row_user_id = t.row_user_id and t1.row_class = t.row_class)
) a
group by row_class;
+---------------------------+-----------+
| sum(row_registration_fee) | row_class |
+---------------------------+-----------+
| 200 | 18 |
| 400 | 19 |
| 0 | 20 |
+---------------------------+-----------+
3 rows in set (0.00 sec)

Answer 2

To get those expected results then it seems to me that you first need to get the unique records per (row_registration_fee, row_class, row_user_id) tupple.
You can use a sub-query with a DISTINCT for that.
Then sum the row_registration_fee.
SELECT
SUM(row_registration_fee) as fee,
row_class
FROM
(
SELECT DISTINCT row_class, row_user_id, row_registration_fee
FROM registration
) q
GROUP BY row_class
Or via a GROUP BY and get the MAX fee.
SELECT
SUM(max_fee) as fee,
row_class
FROM
(
SELECT
row_class, row_user_id,
MAX(row_registration_fee) as max_fee
FROM registration
GROUP BY
row_class, row_user_id
) q
GROUP BY row_class
But to fix your current query, you could remove that WHERE clause.
And then use a CASE WHEN to return NULL on a zero row_registration_fee.
Because a count by value doesn't count the NULL's.
SELECT
COUNT(DISTINCT CASE WHEN row_registration_fee = 0 THEN row_user_id END) * 200 as fee,
row_class
FROM registration
GROUP BY row_class

Answer 3

You need to group by two columns and take sum:
select row_class, sum(row_registration_fee)
from registration
group by row_class, row_user_id

Answer 4

SELECT SUM(row_registration_fee), row_class, row_user_id
FROM registration
WHERE row_registration_fee > 0
GROUP BY row_class, row_user_id;
Yeah, group by two columns you can get per class and user.