MySQL查询按部分获取最大年龄，如果两个或多个具有相同年龄，则返回ID最小的学生

Question

I have a table of students with temporary test values like this:

Table students

+----+-------------+-------+-----------+
| id | section_id  |  age  | name      |
+----+-------------+-------+-----------+
| 1  | 1           | 18    | Justin    |
+----+-------------+-------+-----------+
| 2  | 2           | 14    | Jillian   |
+----+-------------+-------+-----------+
| 3  | 2           | 16    | Cherry    |
+----+-------------+-------+-----------+
| 4  | 3           | 19    | Ronald    |
+----+-------------+-------+-----------+
| 5  | 3           | 21    | Marie     |
+----+-------------+-------+-----------+
| 6  | 3           | 21    | Arthur    |
+----+-------------+-------+-----------+

I want to query the table such that I want to get all the maximum ages of each section. However, if two students have the same age, the table produced will return the student with smallest id.

Return:

+----+------------+-----+--------+
| id | section_id | age | name   |
+----+------------+-----+--------+
| 1  | 1          | 18  | Justin |
+----+------------+-----+--------+
| 3  | 2          | 16  | Cherry |
+----+------------+-----+--------+
| 5  | 3          | 21  | Marie  |
+----+------------+-----+--------+

I tried this query:

SELECT ANY_VALUE(id), ANY_VALUE(section_id), MAX(age), ANY_VALUE(name) FROM
(SELECT id, section_id, age, name FROM students ORDER BY id) as X
GROUP BY section_id

Unfortunately, there are instances that id does not match the age and name.

I have on my end:

sql_mode = only_full_group_by

and I don't have a privilege to edit that, hence the any_value function but I have no idea how to use it.

Answer 1

This will do what you want.

It starts by finding the maximum age per section (including duplicates). Then it joins those results with the minimum id per section (to eliminate duplicates). And finally, select all fields for the matching id and section combinations.

SELECT s3.*
FROM students s3
INNER JOIN (
    SELECT MIN(s2.id) AS id, s2.section_id
    FROM students s2
    INNER JOIN (
        SELECT s1.section_id, MAX(s1.age) AS age
        FROM students s1
        GROUP BY s1.section_id  
    ) s1 USING (section_id, age)
    GROUP BY s2.section_id
) s2 USING (id, section_id);

Working SQL fiddle: https://www.db-fiddle.com/f/aezgAYM6A5KnXykceB7At1/0

Answer 2

I would simply use a correlated subquery:

select s.*
from students s
where s.id = (select s2.id
              from students s2
              where s2.section_id = s.section_id
              order by s2.age desc, s2.id asc
              limit 1
             );

This is pretty much the simplest way to express the logic. And with an index on students(section, age, id) , it should be the most performant as well.