hive grouping sets 等聚合函数

函数说明:

grouping sets
在一个 group by 查询中,根据不同的维度组合进行聚合,等价于将不同维度的 group by 结果集进行 union all
cube
根据 group by 的维度的所有组合进行聚合
rollup
是 cube 的子集,以最左侧的维度为主,从该维度进行层级聚合。

-- grouping sets 
select 
       order_id,
       departure_date,
       count(*) as cnt
  from ord_test
 where order_id=410341346
 group by order_id,
       departure_date
 grouping sets (order_id,(order_id,departure_date))
;

---- 等价于以下
group by order_id
union all
group by order_id,departure_date

-- cube
select 
       order_id,
       departure_date,
       count(*) as cnt
  from ord_test
 where order_id=410341346
 group by order_id,
       departure_date
 with cube
 ;

---- 等价于以下
select count(*) as cnt from ord_test where order_id=410341346
union all
group by order_id
union all
group by departure_date
union all
group by order_id,departure_date

-- rollup
select 
       order_id,
       departure_date,
       count(*) as cnt
  from ord_test
 where order_id=410341346
 group by order_id,
       departure_date
 with rollup
 ;

---- 等价于以下
select count(*) as cnt from ord_test where order_id=410341346
union all
group by order_id
union all
group by order_id,departure_date

结果:grouping_sets, cube, rollup

原文地址:https://www.cnblogs.com/kimbo/p/8445434.html