计算属性总和的 SQL 平均值答案

【问题标题】：SQL Average of Sum of Calculated Attribute计算属性总和的 SQL 平均值
【发布时间】：2015-03-12 00:41:25
【问题描述】：

我试图取一个类别的平均值，其中行按子类别分组并计算总和。父表的主键是子表的分组属性。父表的分组属性既不是主键也不在子表中。

简单表示：

select Category, avg(CalculatedSum)
from ParentTable pt
inner join (
    select Subcategory, sum(Quantity * Price) as 'CalculatedSum'
    from ChildTable
    group by Subcategory
    ) ct
on pt.ID = ct.Subcategory
group by Category

实际的SQL如下：

select c.CU_AGE_RANGE, count(*) as '# of Customers', avg(SumSales) as 'Avg of SumSales', max([Max of SumSales]) as 'Max of SumSales', min([Min of SumSales]) as 'Min of SumSales'
from Customers c
inner join (
    select CUSTOMER_ID, sum(QTY_SOLD * SALES) as SumSales, max(QTY_SOLD*SALES) as 'Max of SumSales', min(QTY_SOLD*SALES) as 'Min of SumSales'
    from Sales
    where (SALES > 0) and (QTY_SOLD > 0) and (COST > 0)
    Group by CUSTOMER_ID
    ) s
on c.CUSTOMER_ID = s.CUSTOMER_ID
group by c.CU_AGE_RANGE

我尝试将 group by 子句更改为 Category (CU_AGE_RANGE) 和 Subcategory (CUSTOMER_ID) 的各种顺序，但总是遇到相同的错误。

错误是表格总是会显示总和的总和（我相信）。我假设这是错误，因为子表中的典型平均值为 250 到 1000，并且 Avg(Sum()) 返回的值大致是每个类别的行数乘以预期的 Sum()。

由于声誉低，我无法发布照片，因此请查看以下逗号分隔结果表：

CU_AGE_RANGE,#_of_Customers,Avg_of_SumSales,Max_of_SumSales,Min_of_SumSales
NULL,125,4261665.306,433460737.7,0.0017
20-29     ,1192,1154040.907,1374037708,0.00025
30-39     ,1902,25429.52329,29426212.64,0.00015
40-49     ,2118,2418.829874,2066725,0.0001
50-59     ,2204,114625.4111,248240261.3,0.00015
60+       ,2135,160156.4341,334617675,0.0005
patrickbig,1,65.5737,12,0.06
Under 19  ,484,1431.262112,92160,0.0001

我试图弄清楚为什么 AVG(SUM()) 返回的似乎是 SUM(SUM())。我目前的预感是，由于 SUM() 是计算条目，因此计算值是根据父表中的分组重新计算的。所以这将是：

希望：

x * y for each row in Child Table
sum(x*y) for each Subcategory
Avg(sum(x/y)) for each Category of Subcategory

QTY_SOLD * SALE for each row in Sales
sum(QTY_SOLD*SALE) for each CUSTOMER_ID
avg(sum(QTY_SOLD*SALE) for each CU_AGE_RANGE group of CUSTOMER_IDs

实际：

x * y for each row in Child Table                  
sum(x * y) for each Subcategory
avg(sum(x * y) for each Category

avg(sum(QTY_SOLD*SALE) for each CU_AGE_RANGE

等于：

sum(QTY_SOLD*SALE) for each CU_AGE_RANGE

如何从当前（类别总和）到期望（按类别平均子类别总和）？

【问题讨论】：

已编辑，我曾尝试格式化，但操作不正确。
@DatMatt 。 . .你的猜测是错误的。查询正在执行您的想法。你可能对CustomerId 有一些奇怪的问题。你能设置一个 SQL Fiddle 吗？
我同意戈登的观点。我找不到你的简单表示有什么问题。

标签： mysql sql join group-by aggregate-functions

【解决方案1】：

您的客户数量有误。您计算的是销售数量，而不是客户数量。更改为count( DISTINCT c.CUSTOMER_ID ) 应该可以解决它。

select c.CU_AGE_RANGE, count( DISTINCT c.CUSTOMER_ID ) as '# of Customers', avg(SumSales) as 'Avg of SumSales', max([Max of SumSales]) as 'Max of SumSales', min([Min of SumSales]) as 'Min of SumSales'
from Customers c
inner join (
    select CUSTOMER_ID, sum(QTY_SOLD * SALES) as SumSales, max(QTY_SOLD*SALES) as 'Max of SumSales', min(QTY_SOLD*SALES) as 'Min of SumSales'
    from Sales
    where (SALES > 0) and (QTY_SOLD > 0) and (COST > 0)
    Group by CUSTOMER_ID
    ) s
on c.CUSTOMER_ID = s.CUSTOMER_ID
group by c.CU_AGE_RANGE

【讨论】：

【解决方案2】：

让我们先考虑子查询：

select Subcategory, sum(Quantity * Price) as 'CalculatedSum'
from ChildTable
group by Subcategory

结果关系的每条记录都表示Subcategory 的聚合。现在，avg(CalculatedSum) 应该产生CalculatedSum 值的平均值。尝试计算sum(CalculatedSum)，看看是否有差异。

【讨论】：