SQL/AWS Athena 和 AWS Quicksight 上的差异分组行

Diff grouped rows on SQL / AWS Athena and AWS Quicksight

我有关于用户跟踪的数据(用户和会话未排序):

user   session    cummulative_time_spent
A      1          2
A      1          5
A      1          10
A      2          3
B      5          1
B      5          200
B      6          5
B      6          6

我想创建花费的时间来区分按会话分组的行。我的预期结果是这样的:

user   session    cummulative_time_spent    duration
A      1          2                         2
A      1          5                         3
A      1          10                        5
A      2          3                         3
B      5          1                         1
B      5          200                       199
B      6          5                         5
B      6          6                         1

我不知道如何在 SQL (AWS Athena) 或 Quicksight 上执行此操作。

您可以使用 lag():

select
    t.*,
    cumulative_time_spent 
        - coalesce(
            lag(cumulative_time_spent) 
                over(partition by user, session order by cumulative_time_spent), 
            0
        ) duration
from mytable

正如 GMB 指出的那样,您想要的功能是 lag()。但是,它有一个简化计算的三参数形式:

select t.*,
       (cumulative_time_spent -
        lag(cumulative_time_spent, 1, cumulative_time_spent) over
           (partition by user, session order by cumulative_time_spent)
       ) as duration
from mytable