Sql 计数案例返回错误的结果

我正在尝试统计在30分钟内和30分钟以上关闭的订单数量

表结构和示例数据

ORDERID  CUSTOMERID  ORDERDATE              CLOSEDDATE               STATUSID
---------------------------------------------------------------------------------
14        1         2018-07-03 11:02:54.000 2018-07-03 13:15:58.000  CLOSED
15        1         2018-07-03 13:22:42.000 NULL                     DISPATCHED
16        1         2018-07-03 13:26:04.000 NULL                     DISPATCHED
17        1         2018-07-03 13:27:57.000 2018-07-03 13:28:28.000  CLOSED
18        1         2018-07-03 17:23:45.000 NULL                     RECEIVED

查询：

SELECT
    OUTLETNAME,
    COUNT(ORDERID) AS ORDERSCOUNT,
    COUNT(DISTINCT(dbo.VW_SALES_SUM.CUSTOMERID)) AS CUSTOMERCOUNT,
    COUNT(CASE WHEN DATEDIFF(mi, dbo.VW_SALES_SUM.ORDERDATE, dbo.VW_SALES_SUM.CLOSEDDATE) < 30 AND dbo.VW_SALES_SUM.STATUSID = 'CLOSED'   
                  THEN 1 
                  ELSE 0
          END) [LESS 30 (CLOSED)],
    COUNT(CASE WHEN DATEDIFF(mi, dbo.VW_SALES_SUM.ORDERDATE, dbo.VW_SALES_SUM.CLOSEDDATE) > 30 AND dbo.VW_SALES_SUM.STATUSID = 'CLOSED'   
                  THEN 1 ELSE 0 
          END) [Greater 30 (CLOSED)]
FROM 
    dbo.VW_SALES_SUM
INNER JOIN 
    dbo.[OUTLET] ON dbo.VW_SALES_SUM.OUTLETCODE = dbo.[OUTLET].CODE
WHERE 
    dbo.VW_SALES_SUM.ORDERDATE BETWEEN '7/3/2018 11:00:00 AM' AND '7/4/2018 02:00:00 AM'
GROUP BY 
    OUTLETNAME, dbo.OUTLET.BRAND
ORDER BY 
    COUNT(ORDERID) DESC

从输出中我们可以看到，它返回了错误的已关闭订单计数，它总是返回订单总数

输出应如下所示：

OUTLETNAME  ORDERSCOUNT CUSTOMERCOUNT   LESS 30 (CLOSED)    Greater 30 (CLOSED)
-------------------------------------------------------------------------
shop1        5           1              0                   2

---

使用SUM代替count

SELECT
    OUTLETNAME,
    COUNT(ORDERID) AS ORDERSCOUNT,
    COUNT(distinct(dbo.VW_SALES_SUM.CUSTOMERID)) as CUSTOMERCOUNT,
    SUM(case when DATEDIFF(mi, dbo.VW_SALES_SUM.ORDERDATE, dbo.VW_SALES_SUM.CLOSEDDATE) < 30 and dbo.VW_SALES_SUM.STATUSID='CLOSED'   then 1 else 0 end) [LESS 30 (CLOSED)],
    SUM(case when DATEDIFF(mi, dbo.VW_SALES_SUM.ORDERDATE,  dbo.VW_SALES_SUM.CLOSEDDATE) > 30 and dbo.VW_SALES_SUM.STATUSID='CLOSED'   then 1 else 0 end) [Greater 30 (CLOSED)]
FROM 
    dbo.VW_SALES_SUM
INNER JOIN 
    dbo.[OUTLET] ON dbo.VW_SALES_SUM.OUTLETCODE = dbo.[OUTLET].CODE
WHERE 
    dbo.VW_SALES_SUM.ORDERDATE BETWEEN '7/3/2018 11:00:00 AM' AND '7/4/2018 02:00:00 AM'
GROUP BY 
    OUTLETNAME, dbo.OUTLET.BRAND
ORDER BY 
    COUNT(ORDERID) DESC

选择
OUTLETNAME，
将（ORDERID）计数为OrderScont，
将（不同的（dbo.VW\u SALES\u SUM.CUSTOMERID））计数为CUSTOMERCOUNT，
总和（当DATEDIFF（mi，dbo.VW_SALES_SUM.ORDERDATE，dbo.VW_SALES_SUM.CLOSEDDATE）小于30且dbo.VW_SALES_SUM.STATUSID='CLOSED'则为1，否则为0结束）[小于30（CLOSED）]，
总和（DATEDIFF（mi，dbo.VW_SALES_SUM.ORDERDATE，dbo.VW_SALES_SUM.CLOSEDDATE）大于30且dbo.VW_SALES_SUM.STATUSID='CLOSED'则1为0结束时的情况）[大于30（已关闭）]
从…起
dbo.VW\u销售金额
内连接
dbo.VW\u SALES\u SUM.OUTLETCODE=dbo[OUTLET].CODE上的dbo[OUTLET]
哪里
dbo.VW_销售_SUM.ORDERDATE介于“2018年7月3日上午11:00:00”和“2018年7月4日上午02:00:00”之间
分组
OUTLETNAME，dbo.OUTLET.BRAND
订购人
计数（订单ID）描述

而不是同时计算1和0。你所要做的就是不要把两者都计算在内

问题是，按值计数不计算空值。
但如果只返回值0或1，则计数的行为与计数（*）没有区别。因为它不返回任何NULL，所以计数可以忽略

所以只要去掉那些

ELSE 0

就可以了

顺便说一句，使用别名可以缩短SQL

SELECT
    OUTLETNAME,
    COUNT(ORDERID) AS ORDERSCOUNT,
    COUNT(DISTINCT(s.CUSTOMERID)) AS CUSTOMERCOUNT,
    COUNT(CASE 
          WHEN DATEDIFF(mi, s.ORDERDATE, s.CLOSEDDATE) < 30 AND s.STATUSID = 'CLOSED'   
          THEN 1
          END) [LESS 30 (CLOSED)],
    COUNT(CASE 
          WHEN DATEDIFF(mi, s.ORDERDATE, s.CLOSEDDATE) > 30 AND s.STATUSID = 'CLOSED'   
          THEN 1
          END) [Greater 30 (CLOSED)]
FROM dbo.VW_SALES_SUM s
JOIN dbo.[OUTLET] o ON o.CODE = s.OUTLETCODE
WHERE 
    s.ORDERDATE BETWEEN '7/3/2018 11:00:00 AM' AND '7/4/2018 02:00:00 AM'
GROUP BY OUTLETNAME, o.BRAND
ORDER BY COUNT(ORDERID) DESC

---

0

和

1

都是非空的，因此生成其中一个的表达式将对每行执行一次

COUNT

。您是否打算改为

SUM（）

？我使用了相同的SUM概念，但在这种情况下，我需要计算此列表中有多少项-0，1，0，0，1？五个，对吗？这就是

COUNT

给你的。如果您想将这些数字相加，请使用

SUM

（或从

案例中取出
ELSE
子句，让它在这种情况下生成
NULL
，这样就不会被计算在内）改为SUM而不是count谢谢Damien的帮助我搜索了这么长时间为什么我的查询出现错误计数。。然后删除这条简单的行
ELSE 0
。。工作！
SELECT
    OUTLETNAME,
    COUNT(ORDERID) AS ORDERSCOUNT,
    COUNT(DISTINCT(s.CUSTOMERID)) AS CUSTOMERCOUNT,
    COUNT(CASE 
          WHEN DATEDIFF(mi, s.ORDERDATE, s.CLOSEDDATE) < 30 AND s.STATUSID = 'CLOSED'   
          THEN 1
          END) [LESS 30 (CLOSED)],
    COUNT(CASE 
          WHEN DATEDIFF(mi, s.ORDERDATE, s.CLOSEDDATE) > 30 AND s.STATUSID = 'CLOSED'   
          THEN 1
          END) [Greater 30 (CLOSED)]
FROM dbo.VW_SALES_SUM s
JOIN dbo.[OUTLET] o ON o.CODE = s.OUTLETCODE
WHERE 
    s.ORDERDATE BETWEEN '7/3/2018 11:00:00 AM' AND '7/4/2018 02:00:00 AM'
GROUP BY OUTLETNAME, o.BRAND
ORDER BY COUNT(ORDERID) DESC

SELECT COUNT(DISTINCT CASE WHEN col1='bar' then col2 END) AS Bars FROM Foo;