将数据透视表中的查询输出与 PostgreSQL 相结合答案

【问题标题】：Combine Output of Queries in Pivot Table with PostgreSQL将数据透视表中的查询输出与 PostgreSQL 相结合
【发布时间】：2021-03-21 15:31:32
【问题描述】：

假设我有一个名为 orders 的表，如下所示：

id	order date	Orders_Wanted	Orders_Given
1	2020-11-29 19:12:44.417	2	6
1	2020-11-29 20:12:44.417	2	6
1	2020-11-30 23:37:28.692	8	2
1	2020-11-30 23:37:28.692	2	6

如何编写一个查询，显示 orders_wanted - orders_given 的计数，按小时分为两列，一列计算正面结果，另一列计算负面结果（请注意，orders_wanted 和 orders_given 是时间，这就是为什么我正在计算 orders_wanted - orders_given）。我还想添加最后一列，用于计算每小时总订单数为正数的百分比 (count_orders_positive/ (count_orders_negative + Count_orders_positive))。

查询的输出如下所示：

week	day	hour	count_orders_positive	count_orders_negative	Percentage_orders_positive
48	7	19	0	1	100%
48	7	20	0	1	100%
49	1	23	1	1	50%

到目前为止，我能够使用这些查询获得最后两个结果，但我不知道如何组合它们。

SELECT  
extract (week from (order_date at time zone 'MST' at time zone 'UTC') ) as "week", 
extract (isodow from (order_date at time zone 'MST' at time zone 'UTC') ) as "day",
extract (hour from (order_date at time zone 'MST' at time zone 'UTC') ) as "hour",
Count (extract (hour from (order_date at time zone 'MST' at time zone 'UTC') )) as 
"count_orders_positive" 
from orders  
WHERE orders_wanted - orders_given >= 0
group by week, day, hour
order by week, day, hour;

week	day	hour	count_orders_positive
49	1	23	1

SELECT
extract (week from (order_date at time zone 'MST' at time zone 'UTC') ) as "week", 
extract (isodow from (order_date at time zone 'MST' at time zone 'UTC') ) as "day",
extract (hour from (order_date at time zone 'MST' at time zone 'UTC') ) as "hour",
Count (extract (hour from (order_date at time zone 'MST' at time zone 'UTC') )) as 
"count_orders_negative" 
from orders  
WHERE orders_wanted - orders_given < 0
group by week, day, hour
order by week, day, hour;

week	day	hour	count_orders_negative
48	7	19	1
48	7	20	1
49	1	23	1

【问题讨论】：

你如何从wanted == 2和given == 6得到count_positive == 1
@s-man 感谢您指出这一点。我的示例表出错了。我现在已经修好了。

标签： sql postgresql datatable count pivot

【解决方案1】：

您可以进行条件聚合。 avg() 可以方便地计算百分比：

select  
    extract (week   from (order_date at time zone 'MST' at time zone 'UTC') ) as "week", 
    extract (isodow from (order_date at time zone 'MST' at time zone 'UTC') ) as "day",
    extract (hour   from (order_date at time zone 'MST' at time zone 'UTC') ) as "hour",
    count(*) filter (where orders_wanted - orders_given >= 0) as count_orders_positive,
    count(*) filter (where orders_wanted - orders_given <  0) as count_orders_negative,
    100 * avg((orders_wanted - orders_given >= 0)::int) as percent_orders_positive
from orders
group by week, day, hour
order by week, day, hour;

【讨论】：