hive 开窗函数

-- 开窗------------------------分组排序---------------------------------
-- row_number() over(partition by ... order by ...)   组内的排名序号 1 2 3 4 5 6 7
-- rank() over(partition by userid order by pv desc) 组内的排名序号 1 2 3 3 5 6 6 8
-- dense_rank over(partition by userid order by pv desc) 组内的排名序号 1 2 3 3 4 5 6 6 7
如果不添加partition by，就是不分组，rank() over(order by pv desc)就是对全部数据，用pv值进行排序

-- 开窗函数：SUM,MAX,MIN, AVG
-- SUM() over(partition by ... order by ... rows between ... and ...)
-- MAX() over(partition by ... order by ... rows between ... and ...)
-- MIN() over(partition by ... order by ... rows between ... and ...)
-- AVG() over(partition by ... order by ... rows between ... and ...)
-- 其中rows between ... and ...

从第一行开始累加到当前行 : 
默认策略： sum(pv) over(partition by userid order by createtime)
使用between：sum(pv) over(partition by userid order by createtime  rows between unbounded preceding and current row)

从向上推3行 累加到当前行 : sum(pv) over(partition by userid order by createtime rows between 3 preceding and current row)
从向上推3行 累加到向后推一行 : sum(pv) over(partition by userid order by createtime rows between 3 preceding and 1 following)
从当前行累加到最后一行 : sum(pv) over(partition by userid order by createtime rows between current row and unbounded following)
从向上推3行 累加到最后一行 : sum(pv) over(partition by userid order by createtime rows between 3 preceding and unbounded following)
从向上推3行，到最后一行 求最大值: max(pv) over(partition by userid order by createtime rows between 3 preceding and unbounded following)
从向上推3行，到最后一行 求最小值 : min(pv) over(partition by userid order by createtime rows between 3 preceding and unbounded following)
从向上推3行，到最后一行 求平均值 : avg(pv) over(partition by userid order by createtime rows between 3 preceding and unbounded following)

-- lag 
将上一行的数据和本行数据放在同一行：LAG(createtime,1,'1970-01-01') OVER(PARTITION BY userid ORDER BY createtime)
将上上一行的数据和本行数据放在同一行：LAG(createtime,2,'1970-01-01') OVER(PARTITION BY userid ORDER BY createtime)

-- lead
将下一行的数据和本行数据放在同一行：LEAD(createtime,1,'1970-01-01') OVER(PARTITION BY userid ORDER BY createtime)
将下下一行的数据和本行数据放在同一行：LEAD(createtime,2,'1970-01-01') OVER(PARTITION BY userid ORDER BY createtime)

标签：...,函数,over,partition,userid,hive,order,开窗,createtime
来源： https://www.cnblogs.com/ljc1212/p/16368150.html