Oracle SQL 分组集返回多于一行答案

【问题标题】：Oracle SQL grouping set returning more than one rowOracle SQL 分组集返回多于一行
【发布时间】：2016-03-11 11:09:24
【问题描述】：

我正在编写一个 Oracle sql 脚本，该脚本显示一个部门位置 (location_id = 1700) 与 所有其他 部门位置 (location_id <> 1700) 的平均值 - 因为它比较两个值，我希望只返回两行。我能够像这样计算出一个查询：

select d.LOCATION_ID, round(avg(e.salary),2) AS "AVG SALARY", count(d.LOCATION_ID) from departments d
join employees e on e.DEPARTMENT_ID = d.DEPARTMENT_ID
where d.location_id = 1700
group by grouping sets(d.LOCATION_ID);

这会为我返回一行：

我的第二个查询返回四行而不是单行（就像我想的那样）：

select round(avg(e.salary),2) AS "AVG SALARY", count(d.LOCATION_ID) from departments d
join employees e on e.DEPARTMENT_ID = d.DEPARTMENT_ID
where d.location_id <> 1700
group by grouping sets(d.LOCATION_ID);

这会返回 4 行，但我希望它只返回 1 行：

我希望解决两个查询，然后将它们合并在一起——显然我需要克服第二个查询的障碍，然后再将它们合并在一起。
有什么想法吗？

【问题讨论】：

标签： sql oracle oracle11g group-by

【解决方案1】：

尝试完全删除 group bys。听起来您只需要平均 where location_id = 1700 和 where location_id 1700 然后 UNION 这两个结果。

select '1700' as "LOCATION", round(avg(e.salary),2) AS "AVG SALARY", count(d.LOCATION_ID) as "COUNT"
from departments d
join employees e on e.DEPARTMENT_ID = d.DEPARTMENT_ID
where d.location_id = 1700

union

select '<>1700' as "LOCATION", round(avg(e.salary),2) AS "AVG SALARY", count(d.LOCATION_ID) as "COUNT"
from departments d
join employees e on e.DEPARTMENT_ID = d.DEPARTMENT_ID
where d.location_id <> 1700

【讨论】：

谢谢，我是在用group by 声明来装逼自己

【解决方案2】：

location_id 应包含在第二个查询的 select 中。现在您可以使用union all 组合结果集。

select d.location_id, round(avg(e.salary),2) AS "AVG SALARY", count(d.LOCATION_ID) 
from departments d
join employees e on e.DEPARTMENT_ID = d.DEPARTMENT_ID
where d.location_id <> 1700
group by d.location_id

但是，这可以使用条件聚合在一个查询中表示为

select
avg(case when d.location_id = 1700 then e.salary else 0 end) 
AS "AVG SALARY for location 1700", 
avg(case when d.location_id <> 1700 then e.salary else 0 end) 
AS "AVG SALARY for locations other than 1700"
from departments d
join employees e on e.DEPARTMENT_ID = d.DEPARTMENT_ID

【讨论】：

【解决方案3】：

您可以考虑使用case 而不是locationid：

select (case when d.lcoation_id = 1700 then 1700 else -1 end) as LocationId,
       round(avg(e.salary), 2) AS "AVG SALARY",
       count(d.LOCATION_ID)
from departments d join
     employees e
     on e.DEPARTMENT_ID = d.DEPARTMENT_ID
group by grouping sets((case when d.lcoation_id = 1700 then 1700 else -1 end));

我不确定您是否需要grouping sets。如果你只想要两行，那么这可能就是你想要的：

select (case when d.lcoation_id = 1700 then 1700 else -1 end) as LocationId,
       round(avg(e.salary), 2) AS "AVG SALARY",
       count(d.LOCATION_ID)
from departments d join
     employees e
     on e.DEPARTMENT_ID = d.DEPARTMENT_ID
group by (case when d.lcoation_id = 1700 then 1700 else -1 end);

【讨论】：