Question

我必须连续6个月提取所有每月交易少于5000的客户名称，然后在第7个月进行3次交易，每次交易20,000次。

客户的所有交易将存储在不同的行中。

示例：考虑客户A，客户的信息将按如下方式存储：

     Name | TransactionDate | Amount
1. CustomerA | 27-08-2015 | 4500
2. CustomerA | 27-09-2015 | 4500
3. CustomerA | 27-10-2015 | 4500
4. CustomerA | 27-11-2015 | 4500
5. CustomerA | 27-12-2015 | 4500
6. CustomerA | 27-01-2016 | 4500
7. CustomerA | 27-02-2016 | 20000
8. CustomerA | 27-02-2016 | 20000
9. CustomerA | 27-02-2016 | 20000

Answer 1

在你指定SQL味道之前，我认为我为T-SQL提供了灵活而体面的解决方案：

1）为了有更简单的查询，我已经定义为持久列来存储月号是一种方便的方法：

create table CustomerTransaction
(
    CustomerName VARCHAR(20),
    TransactionDate DATE,
    Amount NUMERIC(18, 2),
    MonthNo AS DATEPART(yyyy, TransactionDate) * 12 + DATEPART(mm, TransactionDate) - 1 PERSISTED
)

如果无法使用，您可以使用员工日期算术（DATEDIFF），或者精确计算内联。

第一次CTE获取的交易数据包含该组的行号和开始月份编号（客户和付款系列）。
对于每个类别，少量和20K（大额），我从之前的CTE中选择了基于金额的过滤。
对于每个系列，应用计数标准（6小额付款，然后是3大付款）。
根据客户和日期加入小额和大额付款（组月份是最小的组月份 - 1）。

最终查询如下：

declare @SmallAmountsLen INT = 6;
declare @BigAmountsLen INT = 3;
declare @SmallAmountThreshold NUMERIC(18, 2) = 5000
declare @BigAmount NUMERIC(18, 2) = 20000

;with AmountCte AS (
    SELECT CustomerName, TransactionDate, Amount, MonthNo, ROW_NUMBER() OVER (PARTITION BY CustomerName ORDER BY TransactionDate) AS RowNo, 
        MonthNo - ROW_NUMBER() OVER (PARTITION BY CustomerName ORDER BY TransactionDate) AS GroupMonthNo
    FROM CustomerTransaction
),
SmallAmountCte AS (
    SELECT * 
    FROM AmountCte
    WHERE Amount < @SmallAmountThreshold 
),
BigAmountCte AS (
    SELECT * 
    FROM AmountCte
    WHERE Amount = @BigAmount 
),
SmallGroupCte AS (
    select CustomerName, GroupMonthNo 
    from SmallAmountCte
    group by CustomerName, GroupMonthNo 
    having count(1) = @SmallAmountsLen
),
BigGroupCte AS (
    select CustomerName, MonthNo 
    from BigAmountCte
    group by CustomerName, MonthNo 
    having count(1) = @BigAmountsLen
)
select S.*, B.* 
from SmallGroupCte S
    join BigGroupCte B on B.CustomerName = S.CustomerName 
where B.MonthNo = S.GroupMonthNo + @SmallAmountsLen + 1

[编辑]查询，无需计算列

declare @SmallAmountsLen INT = 6;
declare @BigAmountsLen INT = 3;
declare @SmallAmountThreshold NUMERIC(18, 2) = 5000
declare @BigAmount NUMERIC(18, 2) = 20000

;with AmountCte AS (
    SELECT CustomerName, TransactionDate, Amount, DATEPART(yyyy, TransactionDate) * 12 + DATEPART(mm, TransactionDate) - 1 AS MonthNo, 
        ROW_NUMBER() OVER (PARTITION BY CustomerName ORDER BY TransactionDate) AS RowNo, 
        DATEPART(yyyy, TransactionDate) * 12 + DATEPART(mm, TransactionDate) - 1 - ROW_NUMBER() OVER (PARTITION BY CustomerName ORDER BY TransactionDate) AS GroupMonthNo
    FROM CustomerTransaction
),
SmallAmountCte AS (
    SELECT * 
    FROM AmountCte
    WHERE Amount < @SmallAmountThreshold 
),
BigAmountCte AS (
    SELECT * 
    FROM AmountCte
    WHERE Amount = @BigAmount 
),
SmallGroupCte AS (
    select CustomerName, GroupMonthNo 
    from SmallAmountCte
    group by CustomerName, GroupMonthNo 
    having count(1) = @SmallAmountsLen
),
BigGroupCte AS (
    select CustomerName, MonthNo 
    from BigAmountCte
    group by CustomerName, MonthNo 
    having count(1) = @BigAmountsLen
)
select S.*, B.* 
from SmallGroupCte S
    join BigGroupCte B on B.CustomerName = S.CustomerName 
where B.MonthNo = S.GroupMonthNo + @SmallAmountsLen + 1

Answer 2

这是一个更简单的查询来获得结果，我想要的是：

WITH CTE
AS
(
SELECT * FROM
(
SELECT DENSE_RANK() OVER (PARTITION BY Name ORDER BY DATEPART(MONTH,TransactionDate)) AS SrNo,
Name,Amount,DatePart(Month,TransactionDate) AS MonthNo,TransactionDate FROM TransactionTable) AS A
WHERE
(Amount <= 5000 AND SrNo < 7) OR (Amount = 20000 AND SrNo = 7)
)
SELECT A.Name AS Account_Number,A.Transaction_Date FROM
(
SELECT Name,Amount,COUNT(SrNo) As Sr,MAX(TransactionDate) AS Transaction_Date FROM CTE
WHERE SrNo = 7 AND Amount = 20000
GROUP BY Name,Amount
HAVING COUNT(srNo) = 3) AS A
INNER JOIN
(
SELECT Name,COUNT(SrNo) AS Ss FROM CTE
WHERE SrNo < 7 AND Amount <= 5000
GROUP BY Name
HAVING COUNT(srNo) = 6) AS B
ON A.Name = B.Name

让所有客户进行有限的交易

2 个答案: