sql - 如何在 SQL Server 中使用以下方法对数据进行 GROUP BY？缺口和岛屿

Question

DECLARE @Table TABLE 
               (
                    Col_status VARCHAR(5),
                    Col_Location VARCHAR(5),     
                    Col_NUM INT
               )

INSERT INTO @Table (Col_status, Col_Location, Col_NUM)
VALUES ('ON', 'AAA', 1), ('ON', 'AAA', 2),
       ('OFF', 'AAA', 3), ('ON', 'BBB', 4),
       ('ON', 'BBB', 5), ('ON', 'CCC', 6), ('ON', 'BBB', 7),
       ('OFF', 'AAA', 9), ('OFF', 'AAA', 10);

SELECT *
FROM @Table

输出:

Col_status  Col_Location    Col_NUM
----------  ------------    -------
ON          AAA             1
ON          AAA             2
OFF         AAA             3
ON          BBB             4
ON          BBB             5
ON          CCC             6
ON          BBB             7
OFF         AAA             9
OFF         AAA            10

但我期待以下答案；我大部分时间都花在了这上面。

• 按第一列和第二列分组，但应根据第三列值的顺序进行拆分

输出应该是这样的:

Col_status  Col_Location    Min(Col_NUM)    count(Col_NUM)
---------- -------------    -------------   --------------
ON          AAA             1               2
OFF         AAA             3               1
ON          BBB             4               2
ON          CCC             6               1
ON          BBB             7               1
OFF         AAA             9               2

谢谢和问候，凯瑟尔·M

Answer 1

这是一个孤岛和间隙问题。可以这样解决:

select t.Col_status,t.Col_Location, min(t.col_num), count(*)
from
(
   select t.*, row_number() over (partition by Col_status,Col_Location order by Col_NUM) group_rn
   from @Table t
) t
group by t.Col_status,t.Col_Location, t.col_num - t.group_rn

demo

基本思想在t.col_num - t.group_rn中，其中相同序列中的行(具有连续的id)具有相同的差异结果。因此，t.col_num - t.group_rn 允许您隔离这些序列并正确进行分组。