在一个SQL查询中合并两个COUNT

问题描述

我需要查询以显示new_customers X个customers_cancellations

通过此查询，我可以按月获取new_customers：

select count(start_date),to_char(start_date,'MM') as monthNumber,'YY') as yearNumber 
from customer where start_date is not null 
group by to_char(start_date,'MM'),'YY') 
order by yearNumber,monthNumber;

与此同时，我可以按月获得客户取消订单：

select count(cancellation_date),to_char(cancellation_date,'YY') as yearNumber
from customer where cancellation_date is not null 
group by to_char(cancellation_date,monthNumber;

两个查询返回的内容类似：

count|  monthnumber |yearnumber
1    |  1           | 20
7    |  2           | 20
5    |  3           | 20

但是我想要这样的东西：

customer_out_count|customer_in_count| monthnumber   |yearnumber
 0                |1                |   1           | 20
 0                |7                |   2           | 20
 1                |0                |   3           | 20
 0                |1                |   4           | 20
 5                |7                |   5           | 20
 1                |5                |   6           | 20

我已经尝试过另一个查询：

select 'start' as type,count(start_date) as count,'YY') as yearNumber
from customer where start_date is not null
group by to_char(start_date,'YY')
union
select 'cancellation' as type,count(cancellation_date) as counta,'YY') as yearNumber
from customer where cancellation_date is not null
group by to_char(cancellation_date,'YY')
order by yearNumber,monthNumber;

结果没问题：

type                  |newcount         | monthnumber   |yearnumber
 start                |1                |   1           | 20
 cancellation         |1                |   1           | 20
 cancellation         |7                |   2           | 20
 start                |3                |   3           | 20
 cancellation         |1                |   4           | 20
 start                |7                |   5           | 20
 start                |5                |   6           | 20

但是我将需要对代码做一些操作才能实现我所需要的。

如何将这两个查询合并为一个？我正在使用postgreslq。

解决方法

一个选项取消对行的透视，然后聚合。在Postgres中，您可以通过横向联接来表达这一点：

select
    sum(x.customer_in_count)  as customer_in_count,sum(x.customer_out_count) as customer_out_count
    to_char(x.dt,'MM') as monthNumber,to_char(x.dt,'YY') as yearNumber
from customer c
cross join lateral (values
    (c.start_date,1,0),(c.cancellation_date,1)
) as x(dt,customer_in_count,customer_out_count)
where x.dt is not null
group by monthNumber,yearNumber
order by monthNumber,yearNumber