SQL：按连续记录分组第 2 部分

Question

问题 1

我有以下 table 每隔 1 小时显示一个人的位置

Id  EntityID    EntityName  LocationID          Timex   delta
1   1           Mickey      Club house          0300    1
2   1           Mickey      Club house          0400    1
3   1           Mickey      Park                0500    2
4   1           Mickey      Minnies Boutique    0600    3
5   1           Mickey      Minnies Boutique    0700    3
6   1           Mickey      Club house          0800    4
7   1           Mickey      Club house          0900    4
8   1           Mickey      Park                1000    5
9   1           Mickey      Club house          1100    6

每次位置更改时增量增加 +1。我想 return 按照下面的示例按增量分组的聚合。

EntityName  LocationID          StartTime   EndTime
Mickey      Club house          0300        0500
Mickey      Park                0500        0600
Mickey      Minnies Boutique    0600        0800
Mickey      Club house          0800        1000
Mickey      Park                1000        1100
Mickey      Club house          1100        1200

我正在使用从此处获取和改编的以下查询 SQL: Group By on Consecutive Records （效果很好）：

select 
    min(timex) as start_date
    ,end_date
    ,entityid
    ,entityname
    ,locationid
    ,delta
from 
    (
    select 
        s1.timex
        ,(  
         select 
            max(timex)
         from 
            [locationreport2] s2
         where 
            s2.entityid = s1.entityid
            and s2.delta = s1.delta
            and not exists 
                (
                select 
                    null
                from 
                    [dbo].[locationreport2] s3
                where 
                    s3.timex < s2.timex
                    and s3.timex > s1.timex
                    and s3.entityid <> s1.entityid
                    and s3.entityname <> s1.entityname
                    and s3.delta <> s1.delta
                )
            ) as end_date
     , s1.entityid
     , s1.entityname
     , s1.locationid
     , s1.delta
from 
    [dbo].[locationreport2] s1
) Result

group by 
    end_date
    , entityid
    , entityname
    , locationid
    , delta
order by 
    1 asc

但是我不想使用增量（计算和填充它需要付出努力）；相反，我想知道是否有任何方法可以将其计算为部分 / 而运行查询。

我也不介意使用视图。

问题 2

我有以下 table 以 1 小时为间隔显示不同人的位置

Id  EntityID    EntityName  LocationID  Timex   Delta
1   1           Mickey      Club house  0900    1
2   1           Mickey      Club house  1000    1
3   1           Mickey      Park        1100    2
4   2           Donald      Club house  0900    1
5   2           Donald      Park        1000    2
6   2           Donald      Park        1100    2
7   3           Goofy       Park        0900    1
8   3           Goofy       Club house  1000    2
9   3           Goofy       Park        1100    3

我想return按人物和地点分组的汇总。例如

EntityID    EntityName  LocationID  StartTime   EndTime
1           Mickey      Club house  0900    1100
1           Mickey      Park        1100    1200
2           Donald      Club house  0900    1000
2           Donald      Park        1000    1200
3           Goofy       Park        0900    1000
3           Goofy       Club house  1000    1100
3           Goofy       Park        1100    1200

上面的查询（问题1）需要做哪些修改？

Answer 1

听起来像是分析函数的例子。您需要将 EndTime 添加 1 小时，但这取决于数据类型。

SELECT EntityName, LocationID, StartTime, EndTime
  FROM ( SELECT EntityName, LocationID
           ,MIN(Timex) OVER (PARTITION BY EntityID, delta ORDER BY delta) AS StartTime
           ,MAX(Timex) OVER (PARTITION BY EntityID, delta ORDER BY delta) AS EndTime
       FROM locationreport2
    ) x
  GROUP BY EntityName, LocationID, StartTime, EndTime
  ORDER BY EntityName, StartTime

SQL：按连续记录分组第 2 部分

SQL: Group By on Consecutive Records Part 2

sql

sql-server

grouping