Sql server 每种类型客户的销售额百分比

我想找出每种客户类型每周的销售额百分比。我可以看到数学，但我不知道如何编写查询

SELECT 
    c.customerType as 'Customer Type', 
    DATEPART(WEEK, o.orderDate) as 'Week of the year', 
    COUNT(c.customerType)  as 'Number of sales' 
FROM
    [dbo].[Order] o
JOIN 
    Customer c ON c.id = o.customerId
GROUP BY 
    c.customerType, DATEPART(WEEK, o.orderDate)

此查询输出按客户类型分组的每个销售的计数

CustomerType  Week   Number of Sales
------------------------------------
Cash          36      248
Corporate     36       10
Personal      36        5
Cash          37      113
Corporate     37        3
Personal      37        2
Cash          38      136
Corporate     38        7
Personal      38        2
Cash          39      138
Corporate     39        4
Personal      39        3

---

您可以包装查询并使用窗口函数：

select 
    t.*,
    (100.0 * [Number of sales])
        /(sum([Number of sales]) over(partition by [Week of the year]) 
        [Percent of Total Sales]
from (
    select 
        c.customerType as [Customer Type], 
        datepart(week, o.orderDate) as [Week of the year], 
        count(c.customerType) as [Number of sales],     
    from [dbo].[Order] o
    join Customer c ON c.id = o.customerId
    group by c.customerType, datepart(week, o.orderDate), datepart(year, o.orderDate)
) t

注:

• 在SQLServer中，最好使用括号而不是引号来定义标识符（引号通常保留给字符串）

• 100.0

  中的

  .0

  非常重要：它强制SQLServer执行十进制除法（默认情况下，它将使用整数除法，这不是您想要的）

• 我在集团定义中增加了年份；如果您的数据分布在几年内，您可能不希望将不同年份的同一周计算在一起

旁注：SQLServer在混合窗口函数和聚合方面非常灵活。因此，这也可能起作用：

select 
    c.customerType as [Customer Type], 
    datepart(week, o.orderDate) as [Week of the year], 
    count(c.customerType) as [Number of sales],     
    (100.0 * count(c.customerType))
        / (sum(count(c.customerType)) over(partition by datepart(week, o.orderDate)))
        as [Percent of Total Sales]
from [dbo].[Order] o
join Customer c ON c.id = o.customerId
group by c.customerType, datepart(week, o.orderDate), datepart(year, o.orderDate)

---

您可以包装查询并使用窗口函数：

select 
    t.*,
    (100.0 * [Number of sales])
        /(sum([Number of sales]) over(partition by [Week of the year]) 
        [Percent of Total Sales]
from (
    select 
        c.customerType as [Customer Type], 
        datepart(week, o.orderDate) as [Week of the year], 
        count(c.customerType) as [Number of sales],     
    from [dbo].[Order] o
    join Customer c ON c.id = o.customerId
    group by c.customerType, datepart(week, o.orderDate), datepart(year, o.orderDate)
) t

注:

• 在SQLServer中，最好使用括号而不是引号来定义标识符（引号通常保留给字符串）

• 100.0

  中的

  .0

  非常重要：它强制SQLServer执行十进制除法（默认情况下，它将使用整数除法，这不是您想要的）

• 我在集团定义中增加了年份；如果您的数据分布在几年内，您可能不希望将不同年份的同一周计算在一起

旁注：SQLServer在混合窗口函数和聚合方面非常灵活。因此，这也可能起作用：

select 
    c.customerType as [Customer Type], 
    datepart(week, o.orderDate) as [Week of the year], 
    count(c.customerType) as [Number of sales],     
    (100.0 * count(c.customerType))
        / (sum(count(c.customerType)) over(partition by datepart(week, o.orderDate)))
        as [Percent of Total Sales]
from [dbo].[Order] o
join Customer c ON c.id = o.customerId
group by c.customerType, datepart(week, o.orderDate), datepart(year, o.orderDate)

---

您是想要每个客户类型的销售额占客户总销售额的百分比，还是本周总销售额的百分比？请向我们展示您的预期结果。相对于工作日的总销售额，您希望每个客户类型的销售额占客户总销售额的百分比，还是占本周总销售额的百分比？请给我们看看你的预期结果。与本周的总销售额相比