按月分组，在 SQL 中显示重复的月份

Question

我有一个记录仓库数据库事务的表。该表中的信息包括交易发生的地点、交易结束的日期、结束的时间、转移的数量和部门。我试图得到： - 每个部门每月的单位和交易数量如下：

Division Month Transactions Units
1 January 5 10 1 February 10 11 2 February 5 10

我的SQL查询如下：

SELECT
    DISTINCT  tranl.Division, 
    DATENAME(month,tranl.trandate) AS 'Month',
    COUNT(tranl.qty) AS Transactions,
    SUM(tranl.qty) AS Units
FROM translog (nolock) tranl 
WHERE tranl.location = 'stage' 
AND end_tran_date >= '2013-04-01 00:00:00.000' 
GROUP BY end_tran_date, tranl.Division
ORDER BY  tranl.Division

但我得到了不同的结果：

Division Month Transactions Units 1 January 5 10 1 January 3 22 2 February 10 40 2 February 3 12 2 March 1 1 ....

Answer 1

您按 end_tran_date 分组，但您只想按该日期的月份部分分组。以下可能会起作用：

select 
    DISTINCT  tranl.Division, 
    DATENAME(month,tranl.trandate) as 'Month',
     COUNT(tranl.qty) as TRANSACTIONS,
     SUM(tranl.qty) as UNITS
    from translog (nolock) tranl 
    where tranl.location = 'stage' 
   and end_tran_date >= '2013-04-01 00:00:00.000' 
   GROUP BY DATENAME(month,tranl.trandate), tranl.Division
   ORDER BY  tranl.Division

Answer 2

GROUP BY 子句需要指定数据应该按月分组和可能的年份（虽然您没有指定，但我假设您不希望不同年份的交易汇总）：

SQL Server 2012

SELECT
    DISTINCT tranl.division,
    FORMAT(end_tran_date, 'yyyy-MM', 'en-us') AS Month,
    Count(tranl.qty) AS Transactions,
    Sum(tranl.qty) AS Units     
FROM
    translog (nolock) tranl      
WHERE
    tranl.location = 'stage'             
    AND end_tran_date >= '2013-04-01 00:00:00.000'      
GROUP  BY
    tranl.division,
    FORMAT(end_tran_date, 'yyyy-MM', 'en-us')      
ORDER  BY
    tranl.division

SQL Server 2008

SELECT
    DISTINCT tranl.division,
    CONVERT(VARCHAR(7), end_tran_date, 120) AS Month,
    Count(tranl.qty) AS Transactions,
    Sum(tranl.qty) AS Units     
FROM
    translog (nolock) tranl      
WHERE
    tranl.location = 'stage'             
    AND end_tran_date >= '2013-04-01 00:00:00.000'      
GROUP  BY
    tranl.division,
    CONVERT(VARCHAR(7), end_tran_date, 120)      
ORDER  BY
    tranl.division

如果您可以使用，我认为使用 FORMAT 的 SQL Server 2012 语法比用于格式化日期的神秘 CONVERT 语法更易于阅读。