MySQL --- 在同一张表上设置操作 MINUS 和 INTERSECT答案

【问题标题】：MySQL --- set operations MINUS and INTERSECT on the same tableMySQL --- 在同一张表上设置操作 MINUS 和 INTERSECT
【发布时间】：2018-05-02 16:36:15
【问题描述】：

我们需要跟踪随时间变化的集合成员。我可以用下面这个简单的表格来解释我们的问题：

+-----------+------------------+------+-----+---------+----------------+
| Field     | Type             | Null | Key | Default | Extra          |
+-----------+------------------+------+-----+---------+----------------+
| id        | int(10) unsigned | NO   | PRI | NULL    | auto_increment |
| timestamp | varchar(8)       | NO   |     | NULL    |                |
| member    | varchar(4)       | NO   |     | NULL    |                |
+-----------+------------------+------+-----+---------+----------------+

SELECT * 操作的结果如下：

SELECT * FROM My_Table;
+----+-----------+--------+
| id | timestamp | member |
+----+-----------+--------+
|  1 | 20150101  | A      |
|  2 | 20150101  | B      |
|  3 | 20150101  | C      |
|  4 | 20180101  | A      |
|  5 | 20180101  | D      |
|  6 | 20180101  | E      |
+----+-----------+--------+

从逻辑上讲，我们可以使用 SET 操作 MINUS 和 INTERSECT 来了解在一段时间内添加、删除或保留了哪些成员。例如，下面的“逻辑”集合操作给出了在 20150101 和 20180101 之间添加的成员数：

SELECT member FROM my_table WHERE timestamp = "20180101"
MINUS
SELECT member FROM my_table WHERE timestamp = "20150101";

类似地，以下“逻辑”集合操作给出了在 20150101 和 20180101 之间删除的成员数：

SELECT member FROM my_table WHERE timestamp = "20150101"
MINUS
SELECT member FROM my_table WHERE timestamp = "20180101";

以下“逻辑”集合操作给出了 20150101 和 20180101 之间保留的成员数：

SELECT member FROM my_table WHERE timestamp = "20150101"
INTERSECT
SELECT member FROM my_table WHERE timestamp = "20180101";

在 MySQL 中实现这些逻辑集合操作的最优雅的方式是什么？

【问题讨论】：

您需要定义/解释该时间戳值代表什么。
在此上下文中，时间戳表示评估集成员身份的时间
如“成员 X 是时间戳 T 的成员”？
是的，X 是时间戳 T 的成员

标签： mysql set-operations

【解决方案1】：

LEFT JOIN 很可能最能满足您对前两个的需求，第三个是简单的 INNER JOIN。

已添加成员

SELECT t1.member 
FROM my_table AS t1
LEFT JOIN my_table AS t2 ON t1.member = t2.member AND t2.timestamp = "20150101"
WHERE t1.timestamp = "20180101" 
    AND t2.id IS NULL -- this filters results to only those members who did not have an entry in t2
;

会员流失 应该只是能够交换上面的时间戳值。

使用这种通用形式：

SELECT t1.member 
FROM my_table AS t1
LEFT JOIN my_table AS t2 ON t1.member = t2.member AND t2.timestamp = Y
WHERE t1.timestamp = X
    AND t2.id IS NULL -- this filters results to only those members who did not have an entry in t2
;

查询基本上是获取在时间戳 X 处记录的成员，而在时间戳 Y 处未记录为存在的成员。如果 X > Y，那将是获得的成员；如果 X

保留成员

SELECT t1.member 
FROM my_table AS t1
INNER JOIN my_table AS t2 ON t1.member = t2.member 
   AND t2.timestamp = "20150101" -- This condition would be more "properly" 
                                 -- part of the WHERE clause's conditions, 
                                 -- but (unlike the LEFT JOIN queries) whether it 
                                 -- is in the ON or the WHERE will not change the results
WHERE t1.timestamp = "20180101"
;

【讨论】：