Question

我有一张表格，显示每个日期的客户ID列表 - 显示在任何特定日期活跃的客户。因此，每个日期都可以包含同样存在于另一个日期的ID。

bdate            customer_id
2012-01-12       111
2012-01-13       222
2012-01-13       333
2012-01-14       111
2012-01-14       333
2012-01-14       666
2012-01-14       777

我希望编写一个查询来计算两个日期之间唯一ID的总数 - 开始日期是行日期，结束日期是将来的特定日期。

我的查询如下：

select
  bdate,
  count(distinct customer_id) as cts
from users
where bdate between bdate and current_date
group by 1
order by 1

但是这会产生每个日期的唯一用户数量，如下所示：

bdate            customer_id
2012-01-12       1
2012-01-13       2
2012-01-14       4

我想要的结果是（对于开始行日期和2012-01-14之间的用户数量）

bdate            customer_id
2012-01-12       5  - includes (111,222,333,666,777)
2012-01-13       5  - includes (222,333,111,666,777)
2012-01-14       4  - includes (111,333,666,777)

Answer 1

就像@Strawberry所说，你可以像这样建立一个联盟：

select
    t1.bdate,
    count(distinct t2.customer_id) as cts
from users t1
join users t2 on t2.bdate >= t1.bdate
where t1.bdate between t1.bdate and current_date
group by t1.bdate
order by t1.bdate

加入t2可以在特定日期和current_date之间，然后count t2 customer_id为您提供所有用户，就是这样。

SqlFiddle Demo Here

按日期发行的累积计数

1 个答案: