SQL按多列以不同顺序排序后选择奇数行答案

【问题标题】：SQL select odd rows after sorting by multiple columns in different orderSQL按多列以不同顺序排序后选择奇数行
【发布时间】：2021-11-05 16:38:44
【问题描述】：

我有一个学生表（id、分数、部门）和部门表（id、名称）。我正在尝试向查询中添加某些条件。

学校部门

+---------+-------------+
| dept_id | name        |
+---------+-------------+
| 1       | Admin       |
+---------+-------------+
| 2       | Chemistry   |
+---------+-------------+
| 3       | Physics     |
+---------+-------------+
| 4       | Biology     |
+---------+-------------+
| 5       | Mathematics |
+---------+-------------+

学生

+------------+-------+------------+
| student_id | score | department |
+------------+-------+------------+
| 26         | 11    | 4          |
+------------+-------+------------+
| 34         | 11    | 3          |
+------------+-------+------------+
| 76         | 11    | 2          |
+------------+-------+------------+
| 49         | 11    | 1          |
+------------+-------+------------+
| 38         | 11    | 5          |
+------------+-------+------------+

选择所有少于 5 名学生的 school_depts
按学生总分降序排列学校部门。如果出现平局，则获得该部门最多的学生将是第一；如果仍然存在平局，则应首先使用具有最小 dept_id 的部门。
只考虑奇数行，排除偶数行

我的尝试

SELECT * FROM (
SELECT * , ROW_NUMBER() OVER() AS ROW_ID FROM (
SELECT dept_id, name, count(E.student_id) as total_students, SUM(score) as total_score
from Students E
LEFT JOIN school_dept D on dept_id=student_id
group by 1,2
) as O
WHERE total_students<3
order by total_score desc, total_students desc, dept_id asc
) as U
WHERE ROW_ID %2 <>0

预期输出是

name, total_students, total_score
Admin,1,11
Physics1,11
Mathematics,1,11

http://sqlfiddle.com/#!17/248eb8/2

【问题讨论】：

标签： sql sql-server sorting tsql

【解决方案1】：

我认为您的查询中存在连接问题（将 student_id 与 dept_id 连接），并且您选择的学生少于三个（但要求要求少于五个学生）。

请尝试以下方法：

select department as "name", total_students, total_score
from
(
    select *, row_number() over (order by total_score desc, total_students desc, dept_id) row_id
    from
    (
        select sum(score) total_score, count(student_id) total_students, sd.name department, sd.dept_id
        from students s
        join school_dept sd on s.department = sd.dept_id
        group by 3,4
        having count(student_id) < 5
    )t
)tt
where row_id%2 <> 0

请参阅 SQL Fiddle here。

【讨论】：

【解决方案2】：

由于 School_dept 是部门的父表，所以 JOINING 将 School_dept Left 加入学生。因为一个部门可能没有学生，所以左加入是完美的。但是如果考虑那些有学生的部门然后使用 INNER JOIN。

-- SQL SERVER (v2017)
SELECT p.dept_name "Name", p.total_students, p.total_score
FROM (SELECT *
           , ROW_NUMBER() OVER (ORDER BY t.total_score DESC, t.total_students DESC, t.dept_id) row_num
      FROM (SELECT sd.dept_id
                 , MAX(sd.name) dept_name
                 , COUNT(s.student_id) total_students
                 , SUM(s.score) total_score
            FROM School_dept sd
            LEFT JOIN Students s
                   ON sd.dept_id = s.department
            GROUP BY sd.dept_id
            HAVING COUNT(s.student_id) < 5) t) p
WHERE (p.row_num % 2) != 0;

请查看网址https://dbfiddle.uk/?rdbms=sqlserver_2017&fiddle=abf4b40f692d085b98054aa9c9a1dcc7

【讨论】：

@shockwave 请检查这个答案。如果对您有帮助，请告诉我。

【解决方案3】：

我认为下面的查询应该可以工作

select *,
Rank() over  (order by Score desc,TotalStudent desc,dept_id asc) as Serial
from
(
select d.dept_id,d.Name,COUNT(student_id) TotalStudent,sum(Score) Score

from Students s
left join School_dept d on s.department=d.dept_id
where d.dept_id % 2 <> 0
group by d.Name,d.dept_id
) a where TotalStudent < 5

【讨论】：