窗口函数用于对指定的多行数据进行计算并返回计算的结果，与 GROUP BY 的区别在于其仅会将计算结果附加到每一行数据上，而不会对行本身进行合并。
语法
window_function (expression)OVER(
[PARTITIONBY part_key ]
[ORDERBY order_key ]
[ { ROWS| RANGE } BETWEEN frame_start AND frame_end ])
参数说明
参数
说明
window_function
窗口值计算方法，支持通用聚合函数、排序函数和取值函数
PARTITION BY
窗口分区依据
ORDER BY
窗口分区内多行数据排序依据
{ ROWS ｜RANGE } BETWEEN frame_start AND frame_end
窗口帧，即窗口分区内每行数据计算值时使用到的数据范围（行），未指定时代表窗口分区内的所有行
使用示例：
rows between current row and 1 following：当前行及后一行
rows between 1 preceding and current row：当前行及前一行
rows between 1 preceding and 1 following：前一行至后一行（共三行）
rows between current row and unbounded following：当前行及后续所有行
rows between unbounded preceding and current row：当前行及前面的所有行
* |select avg(pv) over(order by analytic_time rows between 2 preceding and current row) as avg_pv_3,pv,analytic_time from (select histogram( cast(__TIMESTAMP__ as timestamp),interval 1 day) as analytic_time, count(*) as pv group by analytic_time order by analytic_time)
