如何在 SQL Server 中按以下方法对数据进行 GROUP BY？差距和岛屿

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

但我希望得到以下答案；我大部分时间都花在这上面。

• 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

感谢和问候，卡西尔 M

Answer 1

这是一个孤岛和缺口问题。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）具有相同的这种差异结果。因此，t.col_num - t.group_rn 允许您隔离这些序列并正确地进行分组。

Answer 2

正如之前的回答所提到的，这是一个差距和孤岛问题。

所以你需要创建一个会生成岛屿的查询，所以这个例子，这里是一个

;WITH cteX
AS(
   SELECT TOP 100 PERCENT
          t.Col_NUM
        , t.Col_Location
        , t.Col_status
        , RN = ROW_NUMBER() OVER (PARTITION BY Col_status,Col_Location ORDER BY Col_NUM) 
   FROM @Table t
   ORDER BY t.Col_NUM
)
SELECT
      X.Col_NUM
    , X.Col_Location
    , X.Col_status
    , X.RN
    , Island = X.Col_NUM - X.RN
FROM
    cteX X
ORDER BY
    X.Col_NUM - X.RN

输出显示群岛

所以使用这个查询你可以得到你想要的结果

;WITH cteX
AS(
   SELECT TOP 100 PERCENT
          t.Col_NUM
        , t.Col_Location
        , t.Col_status
        , RN = ROW_NUMBER() OVER (PARTITION BY Col_status,Col_Location ORDER BY Col_NUM) 
   FROM @Table t
   ORDER BY t.Col_NUM
)

SELECT
      X.Col_status
    , X.Col_Location
    , MIN_Col_Num   = MIN(X.Col_NUM)
    , COUNT_Col_NUM = COUNT(*)
FROM cteX X
GROUP BY
      X.Col_Location
    , X.Col_status
    , (X.Col_NUM - X.RN)
ORDER BY
   MIN_Col_Num,,(X.Col_NUM - X.RN)

输出

Answer 3

试试这个：

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 A.Col_status,A.Col_Location
    ,MIN(A.col_num)[Min(Col_NUM)]
    ,COUNT(1)[count(Col_NUM)]
FROM
(
   SELECT t.*
    ,ROW_NUMBER() OVER (PARTITION BY Col_status,Col_Location ORDER BY Col_NUM) RN
   FROM @Table t
) A
group by A.Col_status,A.Col_Location, A.col_num - A.RN
ORDER BY min(A.col_num)

Answer 4

你能不能只将结果选择到一个临时表中然后这样做：

SELECT Col_Location, 
       Col_Status, 
       MIN (Col_Num), 
       COUNT(*) AS Count(Col_NUM) 
FROM #TMP
GROUP BY Col_Location, 
         Col_Status
ORDER BY 1, 2

Answer 5

在这种情况下，您需要按两列分组 :) 试试这个查询：

SELECT Col_status, Col_Location, MIN(Col_NUM), COUNT(Col_NUM)
FROM @Table
GROUP BY Col_status, Col_Location