Question

我有一个类似下面的场景：

有两个表Transaction和Product

Transaction表有列

Id, Amount ProductId TransactionDate ,DepartmentId
1,  100       100       01/01/2013       1
2,  200       101       02/01/2013       2 and so on....

Product表有列

Id,   Name, ProductType.
100,  AB ,    1
101   C ,     2

我想编写一个输出以下内容的存储过程：

Month Year Department Count(Transactions) Count(Transactions of ProductType1) 
Feb   2012    1              100                        50 
Mar   2012    1              100                        50 
Apr   2012    1              100                        50 
Feb   2012    2              100                        50

我到了这里：

select 
    YEAR(T.TransactionDate) AS [YEAR],
    MONTH(T.TransactionDate) AS [MONTH], 
    Count(T.Id)
from 
    Transaction T 
INNER JOIN 
    Product P ON P.Id = T.ProductId
group by 
    T.DepartmentId, YEAR(T.TransactionDate), MONTH(T.TransactionDate);

输出以下内容：

Month Year Department Count(Transactions)

我想知道我怎么也可以包括：

Count(Transactions of ProductType1)

我也试过这个：

select 
    YEAR(T.TransactionDate) AS [YEAR],
    MONTH(T.TransactionDate) AS [MONTH],    
    Count(T.Id)
    (Select Count(T.Id)
     from Transaction T 
     INNER JOIN Product P ON P.Id = T.ProductId
     where P.Id = 1)
from 
     Transaction T 
INNER JOIN 
     Product P ON P.Id = T.ProductId
group by 
     T.DepartmentId, YEAR(T.TransactionDate), MONTH(T.TransactionDate);

由于group by子句，它给出了productid = 1的Transactions计数的不准确结果

我不想写一个单独的查询..但我想知道是否有一种有效的方法可以让SQL语句在一个查询中返回以下内容？

 Month Year Department Count(Transactions) Count(Transactions of ProductType1)

Answer 1

你真的很接近，你需要添加另一个COUNT，但使用CASE表达式：

SELECT  YEAR(T.TransactionDate) AS [YEAR],
        MONTH(T.TransactionDate) AS [MONTH], 
        COUNT(T.Id) AS Transactions,
        SUM(CASE WHEN P.ProductType = 1 THEN 1 ELSE 0 END) AS TransactionsType1
FROM [Transaction] T 
INNER JOIN Product P 
    ON P.Id = T.ProductId
GROUP BY T.DepartmentId, YEAR(T.TransactionDate), MONTH(T.TransactionDate);

Answer 2

您还可以使用PIVOT函数获取结果：

select month, year, 
  departmentid, totalTransactions, 
  [1] ProductType1,
  [2] ProductType2
from
(
  select month(t.transactiondate) month,
    year(t.transactiondate) year,
    t.departmentid,
    p.productType,
    count(*) over(partition by month(t.transactiondate), 
                                year(t.transactiondate),
                                t.departmentid) totalTransactions
  from [transaction] t
  inner join Product p
    on p.id = t.productid
) s
pivot
(
  count(productType)
  for productType in ([1], [2])
) piv;

请参阅SQL Fiddle with Demo

使用Count＆amp; Sql Query进行查询通过...分组

2 个答案: