MYSQL使用左连接加入3个表

Question

我有这些表格：

pupils: (id_pupil, name, surname, email, user, pass, level, class)
incidents: (id_incidents, date, time, type_incident, comments, id_pupil, id_user, subject, id_trimester)
qualifications: (id_qualification, qualification, date, time, subject, id_pupil, id_user, id_trimester, type_qualification)

我做了这个查询：

SELECT 
    pupils.id_pupil
  , name
  , surname
  , round(avg(qualifications.calificacion),2) as average
  , count(qualifications.calificacion) as count
  , COALESCE(sum(type_incident='miss' and level=1 and class='A' and id_trimester=1),0) as misses
  , COALESCE(sum(type_incident='delay' and level=1 and class='A' and id_trimester=1),0) as delays
  , COALESCE(sum(type_incident='attitude' and level=1 and class='A' and id_trimester=1),0) as attitude
  , COALESCE(sum(type_incident='miss_justif' and level=1 and class='A' and id_trimester=1),0) as misses_justificadas 
FROM 
    pupils 
    left join incidents         ON incidents.id_pupil=pupils.id_pupil
    left join qualifications    ON qualifications.id_pupil=pupils.id_pupil
WHERE 
    level=1
    and class='A' 
    and id_trimester=1 
    and type_qualification='class' 
    and qualifications.id_trimester=incidents.id_trimester
GROUP BY id_pupil

但问题是，我必须得到的正确结果是：

我得到了：

问题在于，增加了班级资格和未命中数（17x6（2个未命中，1个延迟，1个态度和2个misses_just）= 102））并且我不希望这样。我想在第一张图片中显示正确的信息。

你想帮助我吗？

谢谢！

Answer 1

问题是您正在创建的关系会复制大量行。

我个人的做法是：

1. 首先从incidents
获取所需的数据
2. 将“子查询”加入到学生和资格所需的数据

这样的事情：

select 
    p.id_pupil, p.name, p.surname, 
    (case
        when type_incident = 'miss' and level = 1 and class='A' then incident_count 
        else 0
    end) as misses,
    (case
        when type_incident = 'delay' and level = 1 and class='A' then incident_count 
        else 0
    end) as delays,
    (case
        when type_incident = 'attitude' and level = 1 and class='A' then incident_count 
        else 0
    end) as attitude,
    (case
        when type_incident = 'miss_justif' and level = 1 and class='A' then incident_count 
        else 0
    end) as miss_justif
from 
    pupils as p
    left join (
        select id_pupil, class, id_trimester, type_qualification, avg(qualification) as q_avg
        from qualifications
        group by id_pupil, class, id_trimester, type_qualification
    ) as q on p.id_pupil = q.id_pupil
    left join (
        select id_pupil, class, type_incident, level, class, count(id_incidents) as incident_count
        from incidents
        group by id_pupil, id_trimester, incident, level, class
    ) as i on p.id_pupil = i.id_pupil
where ...