in sql how to get the max of sum that is group by two columns

Question

SELECT county, category_name, SUM(bottle_qty*(btl_price-state_btl_cost)) AS profit
FROM sales
GROUP BY county, category_name
ORDER BY profit DESC

I want profit for each county and what category_name is produces the most profit in that county.

So I just want the first row, 8th row and 11th row:

Answer 1

You can use distinct on to solve this greatest-n-per-group problem:

SELECT DISTINCT ON(county)
    county, 
    category_name,
    SUM(bottle_qty*(btl_price-state_btl_cost)) AS profit 
FROM sales 
GROUP BY county, category_name 
ORDER BY county, profit DESC

---

However this does not let you control the ordering of the rows in the resultset. In that case, use row_number() instead:

SELECT *
FROM (
    SELECT
        county, 
        category_name,
        SUM(bottle_qty*(btl_price-state_btl_cost)) AS profit 
        ROW_NUMBER() OVER(
            PARTITION BY county 
            ORDER BY SUM(bottle_qty*(btl_price-state_btl_cost)) DESC
        ) rn
    FROM sales 
    GROUP BY county, category_name 
) t
WHERE rn = 1
ORDER BY profit