Question

我有一个带有userid和timestamp的简单事件表。每一行都是执行事件的用户：

+--------+---------------+
| userid |   timestamp   |
+--------+---------------+
|      3 | 5/25/14 23:30 |
|      2 | 5/25/14 20:57 |
|      1 | 5/25/14 20:18 |
|      2 | 5/25/14 18:49 |
|      2 | 5/25/14 13:41 |
|      1 | 5/25/14 5:47  |
|      3 | 5/24/14 22:34 |
|      1 | 5/24/14 22:07 |
|      2 | 5/24/14 21:53 |
|      1 | 5/24/14 21:42 |
+--------+---------------+

我想添加/选择一个列，其中包含我之前在EACH行中看过用户的次数。所以预期的输出是：

+--------+---------------+-------+
| userid |   timestamp   | count |
+--------+---------------+-------+
|      3 | 5/25/14 23:30 |     2 |
|      2 | 5/25/14 20:57 |     4 |
|      1 | 5/25/14 20:18 |     4 |
|      2 | 5/25/14 18:49 |     3 |
|      2 | 5/25/14 13:41 |     2 |
|      1 | 5/25/14 5:47  |     3 |
|      3 | 5/24/14 22:34 |     1 |
|      1 | 5/24/14 22:07 |     2 |
|      2 | 5/24/14 21:53 |     1 |
|      1 | 5/24/14 21:42 |     1 |
+--------+---------------+-------+

例如，从顶部开始的第三行表示我已经看到用户1总共四次，包括当前的一次。计数为1的底行意味着我只看过一次用户1（这是第一次）。

有没有办法在mysql中进行这种类型的计数？谢谢！

Answer 1

SELECT userid, timestamp,
    (SELECT COUNT(*) FROM myTable t2
    WHERE t2.userid = t1.userid
    AND t2.timestamp <= t1.timestamp) AS user_count
FROM myTable t1

我知道你要求提供一个没有小组的查询，但无论如何都要在这里查询：）

SELECT t1.userid, t1.timestamp, COUNT(*)
FROM myTable t1
JOIN myTable t2 ON t1.userid = t2.userid AND t2.timestamp <= t1.timestamp
GROUP BY t1.userid, t1.timestamp

Answer 2

您也可以使用排名查询

SELECT  
userid,
`timestamp`,rank FROM(
SELECT 
  userid,
 `timestamp`,
@r:= CASE WHEN userid = @g THEN @r+1  ELSE @r:=1 END rank 
,@g:=userid
FROM t,(SELECT @r:=0,@g:=0) t1
ORDER BY userid,timestamp
) t
ORDER BY timestamp DESC

如何计算没有groupby的先前事件的数量

2 个答案:

Demo