大数据之hive：hive新功能之GROUPING SETS，Cube, Rollup

一、GROUPING SETS

1、概述

GROUPING SETS作为GROUP BY的子句,可以简单理解为多条group by语句通过union all把查询结果聚合起来；

2、实战

查看表test_03字段

hive (default)> desc test_03;
OK
col_name	data_type	comment
user_id             	bigint              	                    
device_id           	int                 	                    
os_id               	int                 	                    
app_id              	int                 	                    

--sql1
select device_id,os_id,app_id,count(user_id) from  test_03 group by device_id,os_id,app_id grouping sets((device_id)) 
--等价于sql1
SELECT device_id,null,null,count(user_id) FROM test_03 group by device_id

--sql2
select device_id,os_id,app_id,count(user_id) from  test_03 group by device_id,os_id,app_id grouping sets((device_id,os_id))	
--等价于sql2
SELECT device_id,os_id,null,count(user_id) FROM test_03 group by device_id,os_id
--sql3
select device_id,os_id,app_id,count(user_id) from  test_03 group by device_id,os_id,app_id grouping sets((device_id,os_id),(device_id))
--等价于sql3
SELECT device_id,os_id,null,count(user_id) FROM test_03 group by device_id,os_id 
UNION ALL 
SELECT device_id,null,null,count(user_id) FROM test_03 group by device_id
--sql4
select device_id,os_id,app_id,count(user_id) from  test_03 group by device_id,os_id,app_id grouping sets((device_id),(os_id),(device_id,os_id),())
--等价于sql4	
SELECT device_id,null,null,count(user_id) FROM test_03 group by device_id 
UNION ALL 
SELECT null,os_id,null,count(user_id) FROM test_03 group by os_id 
UNION ALL 
SELECT device_id,os_id,null,count(user_id) FROM test_03 group by device_id,os_id  
UNION ALL 
SELECT null,null,null,count(user_id) FROM test_03

二、Cube

1、概述

cube简称数据魔方，可以实现hive多个任意维度的查询，cube(a,b,c)则首先会对(a,b,c)进行group by，然后依次是(a,b),(a,c),(a),(b,c),(b),©,最后在对全表进行group by，它会统计所选列中值的所有组合的聚合；

2、实战

select device_id,os_id,app_id count(user_id) 
from test_xinyan_reg 
group by device_id,os_id,app_id with cube;

--共有3^0+3^1+3^2+3^3=8种可能,如果维度增大,代码量显著增加
SELECT device_id,null,null ,count(user_id) FROM test_03 group by device_id
UNION ALL
SELECT null,os_id,null,count(user_id) FROM test_xinyan_reg group by os_id
UNION ALL
SELECT device_id,os_id,null,count(user_id) FROM test_xinyan_reg group by device_id,os_id
UNION ALL
SELECT null,app_id,null,count(user_id) FROM test_xinyan_reg group by app_id
UNION ALL
SELECT device_id,null,app_id ,count(user_id) FROM test_xinyan_reg group by device_id,app_id
UNION ALL
SELECT null,os_id,app_id,count(user_id) FROM test_xinyan_reg group by os_id,app_id
UNION ALL
SELECT device_id,os_id,app_id,count(user_id) FROM test_xinyan_reg group by device_id,os_id,app_id
UNION ALL
SELECT null,null,null count(user_id) FROM test_xinyan_reg group by client_version

三、Rollup

1、概述

rollup可以实现从右到做递减多级的统计，显示统计某一层次结构的聚合。

2、实战

--sql1
 select device_id,os_id,app_id,count(user_id) 
from test_xinyan_reg 
group by device_id,os_id,app_id with rollup;
--等价于sql1
select device_id,os_id,app_id,count(user_id) 
from test_xinyan_reg 
group by device_id,os_id,app_id,client_version,from_id 
grouping sets ((device_id,os_id,app_id),(device_id,os_id),(device_id),());

四、Grouping_ID函数

当我们没有统计某一列时，它的值显示为null，这可能与列本身就有null值冲突，这就需要一种方法区分是没有统计还是值本来就是null，grouping_id其实就是所统计各列二进制和；
比如表test_04有两个字段id,type;

id	type
1	1
1	null
2	1
3	2
3	null

SELECT id, type, GROUPING__ID, count(*) from test_04 GROUP BY id, type  grouping sets((id, type),(id)) order by GROUPING__ID;
--结果如下
id	type	grouping__id	_c3
3	2	0	1
2	1	0	1
1	NULL	0	1
1	1	0	1
3	NULL	0	1
3	NULL	1	2
1	NULL	1	2
2	NULL	1	1

grouping_id计算方法
grouping sets 中的每一种粒度，都对应唯一的 grouping__id 值，其计算公式与 group by 的顺序、当前粒度的字段有关。
序号 grouping set 给倒序排列的字段（sex class）赋值 对应的十进制（grouping__id 的值）

1 id 01 1
2 id,type 00 0
这就是上面 grouping sets 的结果中 grouping__id 值的由来。