我有一个带有一些数据的Hive表,我想把它分成15分钟的间隔,然后返回每个间隔的总呼叫持续时间
Hive Table示例:
ID Start End Total Duration
1 1502296261 1502325061 28800
我的输出应显示为:
ID Interval Duration
1 2017-08-09 18:30:00 839
1 2017-08-09 18:45:00 900
1 2017-08-09 19:00:00 900
...
1 2017-08-10 02:15:00 900
1 2017-08-10 02:30:00 61
以有效的方式做到这一点的最佳解决方案是什么?
感谢。
答案 0 :(得分:1)
这是基本的解决方案
显示的时间戳(Interval
)取决于您的系统时区。
with t as (select stack(1,1,1502296261,1502325061) as (`ID`,`Start`,`End`))
select t.`ID` as `ID`
,from_unixtime((t.`Start` div (15*60) + pe.pos)*(15*60)) as `Interval`
, case
when pe.pos = t.`End` div (15*60) - t.`Start` div (15*60)
then t.`End`
else (t.`Start` div (15*60) + pe.pos + 1)*(15*60)
end
- case
when pe.pos = 0
then t.`Start`
else (t.`Start` div (15*60) + pe.pos)*(15*60)
end as `Duration`
from t
lateral view
posexplode(split(space(int(t.`End` div (15*60) - t.`Start` div (15*60))),' ')) pe
;
+----+---------------------+----------+
| id | interval | duration |
+----+---------------------+----------+
| 1 | 2017-08-09 09:30:00 | 839 |
| 1 | 2017-08-09 09:45:00 | 900 |
| 1 | 2017-08-09 10:00:00 | 900 |
| 1 | 2017-08-09 10:15:00 | 900 |
| 1 | 2017-08-09 10:30:00 | 900 |
| 1 | 2017-08-09 10:45:00 | 900 |
| 1 | 2017-08-09 11:00:00 | 900 |
| 1 | 2017-08-09 11:15:00 | 900 |
| 1 | 2017-08-09 11:30:00 | 900 |
| 1 | 2017-08-09 11:45:00 | 900 |
| 1 | 2017-08-09 12:00:00 | 900 |
| 1 | 2017-08-09 12:15:00 | 900 |
| 1 | 2017-08-09 12:30:00 | 900 |
| 1 | 2017-08-09 12:45:00 | 900 |
| 1 | 2017-08-09 13:00:00 | 900 |
| 1 | 2017-08-09 13:15:00 | 900 |
| 1 | 2017-08-09 13:30:00 | 900 |
| 1 | 2017-08-09 13:45:00 | 900 |
| 1 | 2017-08-09 14:00:00 | 900 |
| 1 | 2017-08-09 14:15:00 | 900 |
| 1 | 2017-08-09 14:30:00 | 900 |
| 1 | 2017-08-09 14:45:00 | 900 |
| 1 | 2017-08-09 15:00:00 | 900 |
| 1 | 2017-08-09 15:15:00 | 900 |
| 1 | 2017-08-09 15:30:00 | 900 |
| 1 | 2017-08-09 15:45:00 | 900 |
| 1 | 2017-08-09 16:00:00 | 900 |
| 1 | 2017-08-09 16:15:00 | 900 |
| 1 | 2017-08-09 16:30:00 | 900 |
| 1 | 2017-08-09 16:45:00 | 900 |
| 1 | 2017-08-09 17:00:00 | 900 |
| 1 | 2017-08-09 17:15:00 | 900 |
| 1 | 2017-08-09 17:30:00 | 61 |
+----+---------------------+----------+