SQL - sum (column 1 + column2 + column3 ) and select max value of sum in each group

Question

Here is my table with sample data:

id|group_id|column1|column2|column3| 
--------------------------------------
1    1    |    1        1        0
2    1    |    1        1        1
3    2    |    0        0        0
4    2    |    1        1        1
5    1    |    1        0        0
6    3    |    0        0        0

Expected result set: (The result should show the maximum sum (col 1 + col 2 + col 3) in each group)

id|group_id|column1|column2|column3| 
--------------------------------------
2    1    |    1        1        1
4    2    |    1        1        1
6    3    |    0        0        0

Actual result set: (select *, max(m.column1 + m.column2 + m.column3) as total from my_table m group by m.group_id) which is wrong

id|group_id|column1|column2|column3|total| 
------------------------------------------
1    1    |    1        1        0      3 
3    2    |    0        0        0      3
6    3    |    0        0        0      0  

I'm quite new to SQL, it seems like the query selecting the first id in each group.

what is the best way to get expected result?

Answer 1

Looks like you are trying to get the rows that yield highest sum of 3 columns in each group:

select a.*
from my_table a
join (
  select group_id, max(column1 + column2 + column3) summation
  from my_table
  group by group_id) b on a.group_id = b.group_id and a.column1 + a.column2 + a.column3 = b.summation;

Answer 2

You can do get your desired result ie with a subselect:

First step:

SELECT
    MAX(m1.column1 + m1.column2 + m1.column3)
FROM
    my_table m1
GROUP BY
    m1.group_id

will get you the maximum total per group_id.

Because values in non aggregated columns are indetermined, if this columns contains different values for a group, you can't simply aggregate as you've done, but a subselect using the query from first step will do it:

Complete query

SELECT
     *,
     m.column1 + m.column2 + m.column3 as total
FROM
    my_table m
WHERE
    m.column1 + m.column2 + m.column3 = (
    SELECT
        MAX(m1.column1 + m1.column2 + m1.column3)
    FROM
        my_table m1
    WHERE
        m.group_id = m1.group_id
    GROUP BY
        m1.group_id
);

Demo

Answer 3

You didn't specified exactly which values for column1, column2 and column3 must be returned.

select m.*,n.* from 
  m,
  (select 
    group_id as n_group_id, 
    max(column1+column2+column3) as n_total 
  from m group by group_id) as n 
where m.group_id=n_group_id and (column1+column2+column3)=n_total