在GROUP BY分组中，根据另一列的最大值选择值

Question

I am attempting to build a query that reduces a GROUP BY group to a single row, including a value for a column based on the max value of another column. In this case, I want an item id, total qty ordered and most-used supplier.

I've successfully built a query that sums the qty ordered and groups by item and supplier, yielding:

| id | qty | supplier       |
|  1 | 20  | S&S Activewear |
|  1 | 10  | J&J Textiles   |
|  2 | 5   | AB Footwear    |
|  2 | 10  | CD Shoes       |

and the intended result would be total qty ordered (for all suppliers) and most used supplier, so:

| id | total_qty | most_used_supplier |
|  1 | 30        | S&S Activewear     |
|  2 | 15        | CD Shoes           |

Conceptually, I imagine doing a subquery, grouping the above results by id alone, then sum(qty) and somehow choose the supplier value by ranking the GROUP BY by qty.

I have read many related posts but I am failing to apply any of those methods successfully to this end, including use of ROW_NUMBER and PARTITION_BY.

I am doing this in Elixir with Ecto on a Postgres DB, but to keep it generalized so anyone can respond, I am just looking to understand how this would be done in SQL. Please let me know if I can provide more detail, thank you.

Answer 1

I'm going to suggest multiple subqueries:

select id, sum(qty),
       (select t2.supplier
        from t t2
        where t2.id = t.id
        order by t2.qty desc
        fetch first 1 row only
       ) as supplier
from t
group by id;

This uses standard syntax for returning one row. Your database may have another syntax for the equivalent of fetch first 1 row only .

Answer 2

First find biggest quantities for each id. Then find appropriate suppliers which provide those biggest quantities. Here issue may appear if there are more then one "biggest", and you have to see how to deal with it. Finally, just join it once more to same table, adding appropriate quantity sums.

SELECT item.id, sum(item.qty) total_qty, biggestSupplier.supplier most_used_supplier
from item join
(
    SELECT item.id, supplier       
    from item
    JOIN 
    (
        SELECT id, max(qty) maxqty
        FROM item
        GROUP BY id
    ) maxQtyForId ON item.id = maxQtyForId.id AND item.qty = maxQtyForId.maxqty
) biggestSupplier ON item.id = biggestSupplier.id
group by item.id, biggestSupplier.supplier       

Answer 3

There are several approaches and it sounds like you've played with this one a bit even:

with data as (
    select *,
        row_number() over (partition by id order by qty desc) as rn
    from T
)
select id, sum(qty) as total_qty,
    (select d2.supplier from data d2
        where s2.id = d.id and rn = 1) as most_used_supplier
from data d
group by id;

Answer 4

I Divide the problem in 2. First, finding the max qty and then adding up the qty. Finally, Join the table to get the answers.

SELECT T4.ID, T5.sumQty AS total_qty,T4.supplier AS most_used_supplier
FROM [Test].[dbo].[Test] AS T4 LEFT JOIN
(
    SELECT ID,SUM(QTY) as sumQty
    FROM [Test].[dbo].[Test]
    GROUP BY ID
)AS T5
ON T4.ID = T5.ID
WHERE supplier IN
(
    SELECT supplier 
    FROM [Test].[dbo].[Test] AS T1 LEFT JOIN
        (
          SELECT MAX(qty) AS maxQty, ID
          FROM [Test].[dbo].[Test] AS T
          GROUP BY id
        ) AS T2
    ON T1.ID = T2.ID
    AND T1.qty = T2.maxQty
    WHERE T2.ID IS NOT NULL
)