Question

有人可以解释以下查询之间的差异：

查询1（返回87结果）：

SELECT userId, COUNT(userId) as usercount
FROM `cpnc_PaymentOrder` 
WHERE  created >= UNIX_TIMESTAMP('2011-03-01')
GROUP BY userId
HAVING usercount > 1

查询2（返回177结果）：

SELECT userId, COUNT(userId) as usercount
FROM `cpnc_PaymentOrder` 
WHERE  created >= UNIX_TIMESTAMP('2011-02-01')
GROUP BY userId
HAVING usercount > 1

查询3（返回55结果）：

SELECT userId, COUNT(userId) as usercount
FROM `cpnc_PaymentOrder` 
WHERE  created >= UNIX_TIMESTAMP('2011-02-01')
AND created < UNIX_TIMESTAMP('2011-03-01')
GROUP BY userId
HAVING usercount > 1

现在我认为来自查询2的结果数减去查询1的结果数等于查询3的结果数。但事实并非如此。有人可以解释一下原因吗？

谢谢，乔纳

编辑：

为了澄清，我想写的查询是：

SELECT userId
FROM `cpnc_PaymentOrder` 
WHERE  created >= UNIX_TIMESTAMP('2011-02-01')
AND created < UNIX_TIMESTAMP('2011-03-01')
AND userId "appears in at least one other record from before '2011-03-01'"

Answer 1

因为您要比较组数。

让我们看一个小例子：

user_id | created
      1 | 2011-02-02
      1 | 2011-02-02
      1 | 2011-02-02
      2 | 2011-02-02
      2 | 2011-02-02
      1 | 2011-03-02
      1 | 2011-03-02
      2 | 2011-03-02
      2 | 2011-03-02

此数据集的第一个查询将返回2行，第二个也将返回2行，第3个也将返回2行。

只需删除您的GROUP BY并查看差异（当然，数学不匹配GROUP BY）

Answer 2

查询1包含：WHERE created ＆gt; = UNIX_TIMESTAMP('2011-03-01')
即。日期大于......

查询3包含：WHERE created ＆lt; UNIX_TIMESTAMP('2011-03-01')
即。日期小于......

因此，查询3不是查询1和2的交集。

Answer 3

 user_id | created
       1 | 2011-02-02
       1 | 2011-03-02
       1 | 2011-03-02

Query1将是1个结果，Query2也是1个结果，但是查询3的结果为零，因为您错过了那个时间内只出现过一次的人。

使用UNIX_TIMESTAMP查询中的奇怪差异

3 个答案: