vertica sql：运行总和1小时

Question

我知道有很多关于运行总计的论坛问题，但不知怎的，我不知道如何应用于我的例子。如果有人可以提供帮助，那就太感激了。

我正在尝试按“ID”计算每小时运行的“速度”分区总数。我该如何处理？我正在使用vertica数据库。谢谢！

DATETIME,ID,speed
05-12-15 0:15,1,40
05-12-15 0:30,1,61
05-12-15 0:45,1,66
05-12-15 1:00,1,48
05-12-15 1:15,1,66
05-12-15 1:30,1,50
05-12-15 1:45,1,47
05-12-15 2:00,1,70
05-12-15 2:15,1,42
05-12-15 2:30,1,58
05-12-15 2:45,1,74
05-12-15 3:00,1,43
05-12-15 3:15,1,57
05-12-15 3:30,1,69
05-12-15 3:45,1,52
05-12-15 4:00,1,43
05-12-15 0:15,2,48
05-12-15 0:30,2,73
05-12-15 0:45,2,41
05-12-15 1:00,2,56
05-12-15 1:15,2,65
05-12-15 1:30,2,76
05-12-15 1:45,2,66
05-12-15 2:00,2,49
05-12-15 2:15,2,75
05-12-15 2:30,2,67
05-12-15 2:45,2,76
05-12-15 3:00,2,47
05-12-15 3:15,2,53
05-12-15 3:30,2,76
05-12-15 3:45,2,63
05-12-15 4:00,2,56

编辑： 最终结果应该是这样的：

Answer 1

您应该能够按ID进行分区并使用range between '1 hour' preceding and current row。 RANGE使用ORDER BY中的值来限制窗口框架。

select datetime, 
       id, 
       speed, 
       sum(speed) over (partition by id 
                        order by datetime desc
                        RANGE BETWEEN '1 hour' PRECEDING 
                              AND CURRENT ROW) sum_speed_1hour
from mytable
order by id, datetime

另外，请注意日期时间desc，它基于您的示例是您正在寻找的。

Answer 2

  

您可以尝试以下操作：

select
    TIME_SLICE(datetime, 1, 'HOUR'),
    sum(speed)
from mytable
group by TIME_SLICE(datetime, 1, 'HOUR')
order by TIME_SLICE(datetime, 1, 'HOUR')