Mysql 在一张表上比较两个月以上客户的发展情况

我有一个数据库，每个月都会在其中输入新记录，在这里我可以看到每个提供商的客户数量和客户名称。约有1500家客户和17家供应商。老年客户可以被禁用，也可以每月注册新客户。每个客户只能有一个供应商，但每个月他们都可以更换供应商。我正在尝试执行一个查询，返回每个月谁更改了提供者的客户名称、他们和谁在一起以及他们和谁在一起

我的查询中的所有内容都运行良好，但我希望它能够比较所有月份，而不仅仅是我在表1和表2中写的两个月

我开发了以下代码：

WITH
-- CREATE TABLE 1 FROM A GENERAL DATABASE WITH CURRENT MONTH INFO
t1 AS (
  SELECT 
        Transaction_date,
        Provider,
        Customer_ID,
        Customer_Name,
        CONCAT(Customer_ID, Provider) AS ID1,
        Demand
    FROM MyDataBase
    WHERE YEAR(Transaction_date)=2020
    AND MONTH(Transaction_date)=10
), 

-- CREATE TABLE 2 FROM A GENERAL DATABASE WITH LAST MONTH INFO
t2 AS (
  SELECT 
        Transaction_date AS Transaction_date2,
        Provider AS Provider2,
        Customer_Code AS Customer_ID2,
        Customer_Name AS Customer_Name2,
        CONCAT(Customer_ID, Provider) AS ID2,
        Demand AS Demand2
    FROM MyDataBase
    WHERE YEAR(Transaction_date)=2020
    AND MONTH(Transaction_date)=09
    
),


-- t3: IDENTIFY WHO CHANGED

t3 AS (
    SELECT 
        Transaction_date2,
        Provider2,
        Customer_ID2,
        Customer_Name2,
        Demand2
    FROM t1
    RIGHT JOIN t2
    ON t2.ID2 = t1.ID1
    WHERE demand IS null
),

-- t4: RETURNS THE CUSTOMERS THAT CHANGED OF PRIVIDER IN AN ORGANIZED TABLE

T4 AS (
    SELECT 
        Transacion_date,
        Provider2 AS 'Last_Provider',   
        Provider AS 'Actual_Privder',
        Customer_ID,
        Customer_Name,
        Demand
    FROM t3
        LEFT JOIN t1
        ON t1.Customer_ID = t5.Customer_ID2
        WHERE Demand IS NOT NULL 
        AND Transacion_date IS NOT NULL
    )
    

SELECT * FROM t4

数据库示例

 Month        & Provider & Client & Demand \\
 2020-01-01   & A        & 1      & 50     \\
 2020-01-01   & A        & 2      & 40     \\
 2020-01-01   & B        & 3      & 55     \\
 2020-01-01   & B        & 4      & 70     \\
 2020-02-01   & A        & 1      & 55     \\
 2020-02-01   & B        & 2      & 34     \\
 2020-02-01   & B        & 3      & 67     \\
 2020-02-01   & B        & 4      & 89     \\
 2020-03-01   & A        & 1      & 78     \\
 2020-03-01   & A        & 2      & 89     \\
 2020-03-01   & A        & 3      & 76     \\
 2020-03-01   & B        & 4      & 45    

正如您在示例中所看到的

一月份：有4个客户。 2月：客户2切换到供应商B。 三月：客户2和客户3被转移到提供商A

我正在尝试执行一个查询，返回每个月谁更改了提供者的客户名称、他们和谁在一起以及他们和谁在一起

如果您运行的是MySQL 8.0，那么可以使用lag来获取每个客户以前的提供者。然后，剩下要做的就是筛选提供者更改的月份：

select *
from (
    select t.*,
        lag(provider) over(partition by customer_code order by transaction_date) as previous_provider
    from mytable t
) t
where provider <> previous_provider

在早期版本中，可以使用子查询模拟窗口函数：

select *
from (
    select t.*,
        (
            select t1.provider 
            from mytable t1 
            where t1.customer_code = t.customer_code and t1.transaction_date < t.transaction_date 
            order by t1.transaction_date desc limit 1
        ) as previous_provider
    from mytable t
) t
where provider <> previous_provider

---

谢谢GMB，它工作得很好，查询的持续时间也很快。