Question

我正在使用MySQL数据库，尝试获取已在时间轴内完成某项任务的部门中用户数量的统计信息。

我的问题是：某些用户多次执行任务。我能够构建一个查询，它返回已完成任务的数量和每个组的总用户数，但我只需要为每个用户计算一个“任务”。出于这个原因，当只有一个人完成了足以填满整个部门的要求时，我得到的结果就是“150％的[部门]完成了任务”。

以下是现有查询：

SELECT total.department, total_count, IFNULL(done, 0) as done_count, ROUND((IFNULL(done, 0) / total_count)*100, 2) as percent FROM (SELECT department, COUNT(*) total_count FROM agents GROUP BY department) total LEFT JOIN (SELECT a.department as department, COUNT(*) as done FROM agents a, tasks p WHERE p.task_responses_id IS NOT NULL AND (p.agent1_id = a.id OR p.agent2_id = a.id) GROUP BY a.department) done ON done.department = total.department;

返回一个这样的表（部门名称已清理）：

+------------------+-------------+------------+---------+ | department | total_count | done_count | percent | +------------------+-------------+------------+---------+ | a | 2 | 0 | 0.00 | | b | 10 | 1 | 10.00 | | c | 2 | 0 | 0.00 | | d | 1 | 0 | 0.00 | | e | 2 | 2 | 100.00 | | f | 1 | 0 | 0.00 | | g | 3 | 6 | 200.00 | | h | 4 | 0 | 0.00 | | i | 4 | 1 | 25.00 | +------------------+-------------+------------+---------+

如你所见，部门“g”已经完成了＆count; total_count由于该部门的一个人多次这样做。我需要使用看起来像这样的任务表：

+-----+----------------+-----------+-----------+-----------------------+---------------------+------+ | id | reservation_id | agent1_id | agent2_id | task_responses_id | last_contact | dnc | +-----+----------------+-----------+-----------+-----------------------+---------------------+------+ | 128 | 6457633 | 9 | NULL | 24 | 2015-10-06 00:00:00 | 1 | | 130 | 6799659 | 10 | NULL | 25 | 2015-10-06 00:00:00 | NULL | | 145 | 7004981 | 36 | NULL | 28 | 2015-10-08 00:00:00 | NULL | | 150 | 7091836 | 36 | NULL | 29 | 2015-10-08 00:00:00 | NULL | | 152 | 7128330 | 36 | NULL | 30 | 2015-10-08 00:00:00 | NULL | | 155 | 7165876 | 16 | NULL | 35 | 2015-10-08 00:00:00 | NULL | | 166 | 7308234 | 36 | NULL | 31 | 2015-10-08 00:00:00 | NULL | | 171 | 7333373 | 36 | NULL | 33 | 2015-10-08 00:00:00 | NULL | | 173 | 7408857 | 37 | NULL | 34 | 2015-10-08 00:00:00 | NULL | +-----+----------------+-----------+-----------+-----------------------+---------------------+------+

如果我们已经为给定的代理ID检索了一行，我想不要抓住任何其他人的ID。

非常感谢你的帮助！我很乐意澄清您可能遇到的任何问题。

Answer 1

我认为这可以通过将第3行中的“count（*）”替换为“count（distinct a.id）”来实现

这样，如果同一个代理ID不止一次存在，它将只计算一次。

所以查询看起来像这样：

SELECT total.department, total_count, IFNULL(done, 0) as done_count, ROUND((IFNULL(done, 0) / total_count)*100, 2) as percent 
FROM (SELECT department, COUNT(*) total_count FROM agents GROUP BY department) total 
LEFT JOIN (SELECT a.department as department, COUNT(distinct a.id) as done FROM agents a, tasks p WHERE p.task_responses_id IS NOT NULL AND (p.agent1_id = a.id OR p.agent2_id = a.id) 
GROUP BY a.department) done ON done.department = total.department;

Answer 2

要计算同一查询中每个部门的代理数量与完成任务的数量，您可以在选择列表中使用子查询，但这也不会执行。相反，我推荐以下内容，它更复杂但性能最佳：

SELECT  d.department, count(*) as dept_count, sum(d.done) as done_count
FROM    (SELECT *,
                (CASE WHEN EXISTS(
                        SELECT * FROM tasks
                        WHERE    (agents.id = tasks.agent1_id OR agents.id = tasks.agent2_id)
                          AND    tasks.task_responses_id IS NOT NULL
                    ) THEN 1 ELSE 0 END
                ) as done
        FROM    agents
        ) as d
GROUP BY department;

此版本使用代理表上的内部查询来添加＆＃34;完成＆＃34;如果该代理符合条件，则值为1的列，否则为0。外部查询计算所有行，但也总结1到1的数量以获取done_count。

Answer 3

您应该使用EXISTS子查询（这也称为半连接）。您想要计算某些条件所适用的用户数。我没有完整的架构，但看起来应该做你想做的事情：

SELECT department, count(*) AS done_count
FROM   agents
WHERE EXISTS(
    SELECT * FROM tasks
    WHERE   (agents.id = tasks.agent1_id OR agents.id = tasks.agent2_id)
      AND   tasks.task_responses_id IS NOT NULL
)
GROUP BY department;

此查询几乎完全符合您在第一段中所要求的内容。通过避免使用LEFT JOIN和DISTINCT运算符，您可以为DBMS提供构建合理查询的机会，该查询的运行时间不会超过其所需的时间。

MySQL返回与给定id相关的第一行

3 个答案: