Hive分析窗口函数之SUM，AVG，MIN和MAX

Hive中提供了很多的分析函数，用于完成负责的统计分析。

本文先介绍SUM、AVG、MIN、MAX这四个函数。

环境信息：

Hive版本为apache-hive-0.14.0-bin

Hadoop版本为hadoop-2.6.0

Tez版本为tez-0.7.0

构造数据：

P088888888888,2016-02-10,1

P088888888888,2016-02-11,3

P088888888888,2016-02-12,1

P088888888888,2016-02-13,9

P088888888888,2016-02-14,3

P088888888888,2016-02-15,12

P088888888888,2016-02-16,3

创建表：

hive (hiveinaction)> create table windows_func

                   >(

                  >     polno string,

                  >     createtime string,

                  >     pnum int   

                   >)

                   >ROW FORMAT DELIMITED

                   >FIELDS TERMINATED BY ','

                   >stored as textfile;

导入数据到表中：

load data local inpath '/home/hadoop/testhivedata/windows_func.txt' into table windows_func;

测试：

SELECT polno,

       createtime,

       pnum,

       SUM(pnum) OVER(PARTITION BY polno ORDERBY createtime) AS pnum1, --默认为从起点到当前行

       SUM(pnum) OVER(PARTITION BY polno ORDERBY createtime ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS pnum2,  --从起点到当前行

       SUM(pnum) OVER(PARTITION BY polno) ASpnum3, --分组内所有行

       SUM(pnum) OVER(PARTITION BY polno ORDERBY createtime ROWS BETWEEN 3 PRECEDING AND CURRENT ROW) AS pnum4,  --当前行+往前3行(当前行的值+前面三行的值)

       SUM(pnum) OVER(PARTITION BY polno ORDERBY createtime ROWS BETWEEN 3 PRECEDING AND 1 FOLLOWING) AS pnum5,  --当前行+往前3行+往后1行

       SUM(pnum) OVER(PARTITION BY polno ORDERBY createtime ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING) AS pnum6 ---当前行+往后所有行

FROM windows_func;

结果：

polno	createtime	pnum	pnum1	pnum2	pnum3	pnum4	pnum5	pnum6
P088888888888	2016/2/10	1	1	1	32	1	4	32
P088888888888	2016/2/11	3	4	4	32	4	5	31
P088888888888	2016/2/12	1	5	5	32	5	14	28
P088888888888	2016/2/13	9	14	14	32	14	17	27
P088888888888	2016/2/14	3	17	17	32	16	28	18
P088888888888	2016/2/15	12	29	29	32	25	28	15
P088888888888	2016/2/16	3	32	32	32	27	27	3

注释：

1.   如果不指定ROWS BETWEEN,默认为从起点到当前行;

2.   如果不指定ORDER BY，则将分组内所有值累加;

理解ROWS BETWEEN含义,也叫做WINDOW子句：
PRECEDING：往前
FOLLOWING：往后
CURRENT ROW：当前行
UNBOUNDED：起点，UNBOUNDED PRECEDING表示从前面的起点， UNBOUNDED FOLLOWING：表示到后面的终点
其他AVG，MIN，MAX，和SUM用法一样。

 

演示AVG环境：

SELECT polno,

       createtime,

       pnum,

       AVG(pnum) OVER(PARTITION BY polno ORDER BY createtime) AS pnum1, --默认为从起点到当前行

       AVG(pnum) OVER(PARTITION BY polno ORDER BY createtime ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS pnum2,  --从起点到当前行

       AVG(pnum) OVER(PARTITION BY polno) AS pnum3, --分组内所有行

       AVG(pnum) OVER(PARTITION BY polno ORDER BY createtime ROWS BETWEEN 3 PRECEDING AND CURRENT ROW) AS pnum4,  --当前行+往前3行(当前行的值+前面三行的值)

       AVG(pnum) OVER(PARTITION BYpolno ORDER BY createtime ROWS BETWEEN 3 PRECEDING AND 1 FOLLOWING) AS pnum5,  --当前行+往前3行+往后1行

       AVG(pnum) OVER(PARTITION BYpolno ORDER BY createtime ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING) AS pnum6 ---当前行+往后所有行

FROM windows_func;

结果：

polno	createtime	pnum	pnum1	pnum2	pnum3	pnum4	pnum5	pnum6
P088888888888	2016/2/10	1	1	1	4.57142857	1	2	4.5714286
P088888888888	2016/2/11	3	2	2	4.57142857	2	1.666667	5.1666667
P088888888888	2016/2/12	1	1.66667	1.6667	4.57142857	1.666667	3.5	5.6
P088888888888	2016/2/13	9	3.5	3.5	4.57142857	3.5	3.4	6.75
P088888888888	2016/2/14	3	3.4	3.4	4.57142857	4	5.6	6
P088888888888	2016/2/15	12	4.83333	4.8333	4.57142857	6.25	5.6	7.5
P088888888888	2016/2/16	3	4.57143	4.5714	4.57142857	6.75	6.75	3

 

其他类似的函数就不举例了。