我遇到这样的情况，我们的系统为系统上的最终用户创建了多个重叠的行用于登录活动。我不知道为什么会这样做，但确实如此。我将在下面添加一些行作为示例：
RN2 UserRegistryID LoginTime LogoutTime FinalLoginTime FinalLogoutTime
-------------------- -------------- ----------------------- ----------------------- -------------- ---------------
1 x89889 2018-05-15 12:56:30.000 2018-05-15 13:08:24.873
2 x89889 2018-06-26 09:08:59.000 2018-06-26 09:22:24.003
3 x89889 2018-06-26 09:22:58.000 2018-06-26 09:51:02.057
4 x89889 2018-11-09 12:50:58.000 2018-11-09 13:33:56.250
5 x89889 2019-02-12 13:16:17.000 2019-02-12 13:18:39.293
6 x89889 2019-02-12 13:19:38.000 2019-02-12 13:38:02.627
7 x89889 2019-02-19 13:52:00.000 2019-02-19 14:22:18.000
8 x89889 2019-02-19 14:23:34.000 2019-02-19 15:24:02.000
9 x89889 2019-03-03 13:20:52.000 2019-03-03 13:29:50.000
10 x89889 2019-03-03 13:30:25.000 2019-03-03 15:55:02.247
11 x89889 2019-06-21 12:19:35.000 2019-06-21 12:34:01.103
12 x89889 2019-09-17 07:55:06.000 2019-09-17 09:08:26.007
13 x89889 2019-09-19 20:22:40.000 2019-09-19 20:23:01.723
14 x89889 2019-09-21 23:21:43.000 2019-09-22 00:50:10.867
15 x89889 2019-09-23 00:16:50.000 2019-09-23 00:55:35.183
16 x89889 2019-10-13 22:35:43.000 2019-10-13 23:21:34.000
17 x89889 2019-10-13 23:16:29.000 2019-10-14 00:18:55.000
18 x89889 2019-10-14 00:16:09.000 2019-10-14 00:47:25.003
19 x89889 2019-10-14 12:24:24.000 2019-10-14 12:45:19.000
20 x89889 2020-01-07 15:07:42.000 2020-01-07 15:28:49.093
21 x89889 2020-01-29 14:29:41.000 2020-01-29 15:05:08.223
22 x89889 2020-02-10 12:31:04.000 2020-02-10 12:37:36.343
23 x89889 2020-03-17 19:10:31.000 2020-03-17 19:52:37.003
24 x89889 2020-03-24 15:23:47.000 2020-03-24 15:54:15.000
25 x89889 2020-03-24 16:31:42.000 2020-03-24 16:46:56.000
26 x89889 2020-03-25 21:04:43.000 2020-03-25 21:27:11.000
27 x89889 2020-03-25 21:45:56.000 2020-03-25 22:50:19.003
28 x89889 2020-03-26 01:39:16.000 2020-03-27 09:30:09.003
29 x89889 2020-03-26 18:15:36.000 2020-03-26 18:35:50.000
30 x89889 2020-04-09 18:47:32.000 2020-04-09 19:06:02.000
31 x89889 2020-04-16 19:13:57.000 2020-04-16 20:02:04.000
32 x89889 2020-04-24 09:13:07.000 2020-04-24 09:33:16.000
因此，RN2列按事件发生的顺序对其进行排序，并且按UserRegistryID定义的顺序对每个用户进行分区和递增。如您所见，第二行的登录时间早于第一行的注销时间。第三行也是如此。因此，通过查看此内容，您可以推断出这应该被识别为一个“会话”，从第一行的LoginTime和最后一行的LogoutTime开始。我一直在尝试找出解决方法，并尝试了许多方法，但都没有成功。有谁知道我将如何实现这一目标？
非常感谢您提前提供帮助。
这是一个孤岛问题。这是一种使用lag()和累积sum()定义连续重叠行的组的方法，然后可以对其进行汇总：
select
userRegistryID,
min(rn2) min_rn2,
max(rn2) max_rn2,
min(loginTime) minLoginTime,
max(logoutTime) maxLogoutTime,
count(*) no_records
from (
select
t.*,
sum(case when loginTime <= lagLogoutTime then 0 else 1 end)
over(partition by userRegistryID order by rn2) grp
from (
select
t.*,
lag(logoutTime) over(partition by userRegistryID order by rn2) lagLogoutTime
from mytable t
) t
) t
group by userRegistryID, grp
order by userRegistryID, minLoginTime
或者，如果您不想汇总行，而是将每个会话的开始和结束日期添加到每一行，则可以执行：
select
rn2,
userRegistryID,
loginTime,
logoutTime,
min(loginTime) over(partition by userRegistryID, grp) finalLoginTime,
max(logoutTime) over(partition by userRegistryID, grp) finalLogoutTime
from (
select
t.*,
sum(case when loginTime <= lagLogoutTime then 0 else 1 end)
over(partition by userRegistryID order by rn2) grp
from (
select
t.*,
lag(logoutTime) over(partition by userRegistryID order by rn2) lagLogoutTime
from mytable t
) t
) t
order by userRegistryID, rn2