Question

我想根据 customer_code列和品牌列<更新“品牌A”，“品牌B”和“品牌C”列，其结果为“true”或“false” / b>。

例如，表市场显示以下内容：我有一个顾客买了A和B品牌：

Customer_code Brand Brand A Brand B Brand C 1234567 A NULL NULL NULL 1234567 B NULL NULL NULL Customer_code Brand Brand A Brand B Brand C 1234567 A True True False 1234567 B True True False

由于我有大量数据，所以有什么方法可以解决这个问题吗？

非常感谢你！

Answer 1

以这种方式存储数据有很多或冗余：

Customer_code   Brand   Brand A Brand B Brand C
1234567           A     True    True    False
1234567           B     True    True    False

存储它的正常方法就是：

Customer_code   Brand 
1234567           A    
1234567           B

由于您的真/假列基于前两个数据，因此存储它们是一个坏主意，因为您有另外的任务来保持其准确性。

想象一下，我向你的表中插入了一个新行，即使我为每个品牌插入了正确的true / false值，它仍会使前两行错误：

Customer_code   Brand   Brand A Brand B Brand C
1234567           A     True    True    False
1234567           B     True    True    False
1234567           C     True    True    True

正如我所说，我建议只存储customer_code和品牌，其余的可以从这两列中计算出来。最后，我还建议使用BIT DataType来表示布尔值，而不是将true或false存储为文本（SQL Server中没有布尔类型）。所以你可以创建一个视图：

SELECT  pvt.Customer_Code,
        [Brand A] = CAST(CASE WHEN pvt.[A] > 0 THEN 1 ELSE 0 END AS BIT),
        [Brand B] = CAST(CASE WHEN pvt.[B] > 0 THEN 1 ELSE 0 END AS BIT),
        [Brand C] = CAST(CASE WHEN pvt.[C] > 0 THEN 1 ELSE 0 END AS BIT)
FROM    (   SELECT  Customer_Code, Brand, Value = 1
            FROM    T
        ) T
        PIVOT
        (   COUNT(Value)
            FOR Brand IN ([A], [B], [C])
        ) pvt;

这给出了以下结果：

Customer_code | Brand A | Brand B | Brand C
--------------+---------+---------+----------
1234567       |     1   |     1   |     0

<强> Example on SQL Fiddle

它没有品牌专栏，但这是多余的，因为品牌是每个品牌的位栏所暗示的，因此包括品牌只会导致重复的行。

最后，要真正回答您的问题，您可以使用以下内容实际进行更新：

WITH PVT AS
(   SELECT  Customer_Code, [A], [B], [C]
    FROM    (   SELECT  Customer_Code, Brand, Value = 1
                FROM    T
            ) T
            PIVOT
            (   COUNT(Value)
                FOR Brand IN ([A], [B], [C])
            ) pvt
)
MERGE T
USING PVT
    ON PVT.Customer_Code = T.Customer_Code
WHEN MATCHED THEN UPDATE
    SET [Brand A] = CASE WHEN pvt.A > 0 THEN 'True' ELSE 'False' END,
        [Brand B] = CASE WHEN pvt.B > 0 THEN 'True' ELSE 'False' END,
        [Brand C] = CASE WHEN pvt.C > 0 THEN 'True' ELSE 'False' END;

<强> Example on SQL Fiddle

或SQL Servier特定的UPDATE / FROM语法：

WITH PVT AS
(   SELECT  Customer_Code, [A], [B], [C]
    FROM    (   SELECT  Customer_Code, Brand, Value = 1
                FROM    T
            ) T
            PIVOT
            (   COUNT(Value)
                FOR Brand IN ([A], [B], [C])
            ) pvt
)
UPDATE  T
SET     [Brand A] = CASE WHEN pvt.A > 0 THEN 'True' ELSE 'False' END,
        [Brand B] = CASE WHEN pvt.B > 0 THEN 'True' ELSE 'False' END,
        [Brand C] = CASE WHEN pvt.C > 0 THEN 'True' ELSE 'False' END
FROM    T
        INNER JOIN PVT
            ON PVT.Customer_Code = T.Customer_Code;

<强> Example on SQL Fiddle

Answer 2

使用子查询：

UPDATE m
SET Brand_A = CASE WHEN EXISTS(SELECT 1 FROM dbo.Market m2
                               WHERE m.Customer_code = m2.Customer_code
                               AND   m2.Brand = 'A') THEN 1 ELSE 0 END,
    Brand_B = CASE WHEN EXISTS(SELECT 1 FROM dbo.Market m2
                               WHERE m.Customer_code = m2.Customer_code
                               AND   m2.Brand = 'B') THEN 1 ELSE 0 END,
    Brand_C = CASE WHEN EXISTS(SELECT 1 FROM dbo.Market m2
                               WHERE m.Customer_code = m2.Customer_code
                               AND   m2.Brand = 'C') THEN 1 ELSE 0 END
FROM dbo.Market m

Demo

Answer 3

create table market
(customer_code varchar(200),
brand char(1),
brand_a int,
brand_b int,
brand_c int
);

insert into market values('1234567','A',0,0,0);
insert into market values('1234567','B',0,0,0);

UPDATE market
SET brand_a = a, brand_b = b, brand_c = c
FROM
(
 SELECT customer_code c_code,
 MAX(CASE WHEN brand = 'A' THEN 1 ELSE NULL END) over(PARTITION by customer_code) a,
 MAX(CASE WHEN brand = 'B' THEN 1 ELSE NULL END) over(PARTITION by customer_code) b,
 MAX(CASE WHEN brand = 'C' THEN 1 ELSE NULL END) over(PARTITION by customer_code) c
 FROM market
) aa

WHERE customer_code = c_code;

另外一件事就是删除不需要的列，只将客户和品牌存储在数据库中，稍后再使用查询：

    SELECT 
    customer_code,
    MAX(CASE WHEN brand='A' THEN 1 ELSE 0 END) AS brand_a,
    MAX(CASE WHEN brand='B' THEN 1 ELSE 0 END) AS brand_b,
    MAX(CASE WHEN brand='C' THEN 1 ELSE 0 END) AS brand_c
    FROM market
    GROUP BY customer_code;

使用true或false更新sql列

3 个答案: