Question

我试图弄清楚如何用SQL做到这一点。我在客户表中有一个带有以下各列的表-（Customer_Id，性别，生日）。问题是-我需要找到Gender的最早和最新出生的人。本质上是不同组的最小值和最大值。

123 M   2017-07-05 00:00:00.000
345 M   2016-08-01 00:00:00.000
555 F   2012-01-09 00:00:00.000
567 F   2015-02-07 00:00:00.000
789 F   2013-01-02 00:00:00.000
111 F   2000-01-01 00:00:00.000
188 M   2008-09-01 00:00:00.000

结果集应如下所示

188     M       2008-09-01 00:00:00.000
123     M       2017-07-05 00:00:00.000
111     F       2000-01-01 00:00:00.000
567     F       2015-02-07 00:00:00.000

我可以做4个UNION并以这种方式找出答案，但这效率不高。

这是我想出的，但这也不起作用。如何在一个查询中同时为MAX组做到这一点？

    select a.Customer_id, a.gender, b.min_birthday
    from(
    select gender, min(birthday) min_birthday
    from Sales..Customer group by gender) b join Sales..Customer a on b.gender = a.gender 
    and b.min_birthday = a.birthday

Answer 1

一种方法使用窗口函数：

select customer_id, gender, birthday
from (select c.*,
             row_number() over (partition by gender order by birthday) as seqnum_asc,
             row_number() over (partition by gender order by birthday desc) as seqnum_desc
      from customer c
     ) c
where 1 in (seqnum_asc, seqnum_desc);

如果要联系，请使用rank()代替row_number()。

也就是说，使用(gender, birthday)和(gender, birthday desc)上的索引（如果优化程序有所改进，可能不再需要两个索引），union all方法应该表现得很好：

select c.*
from ((select top (1) c.*
       from customer c
       where gender = 'M'
       order by birthday
      ) union all
      (select top (1) c.*
       from customer c
       where gender = 'F'
       order by birthday
      ) union all
      (select top (1) c.*
       from customer c
       where gender = 'M'
       order by birthday desc
      ) union all
      (select top (1) c.*
       from customer c
       where gender = 'F'
       order by birthday desc
      )
     ) c;

Answer 2

实际上一个UNION ALL就足够了：）

select gender, min(birthday) birthday, 'MIN' Aggregate
from Sales..Customer group by gender
union all
select gender, max(birthday), 'MAX'
from Sales..Customer group by gender

但是更好的痤疮应该是：

select gender, min(birthday), max(birthday)
from Sales..Customer group by gender

但是结果将与期望的略有不同。

Answer 3

您可以使用“不存在”来做到这一点：

select c.* from Customer c
where not exists (
  select 1 from Customer
  where gender = c.gender and birthday < c.birthday
) or not exists (
  select 1 from Customer
  where gender = c.gender and birthday > c.birthday
)
order by c.gender, c.birthday

请参见demo。
结果：

> id  | gender | birthday           
> :-- | :----- | :------------------
> 111 | F      | 01/01/2000 00:00:00
> 567 | F      | 07/02/2015 00:00:00
> 188 | M      | 01/09/2008 00:00:00
> 123 | M      | 05/07/2017 00:00:00

如何使用SQL查找每个组的最小值和最大值？

3 个答案: