我有4张桌子：
Entries:
id
year
day
AnalysisText:
id
text
entries_id
AnalysisType:
id
type
AnalysisTextTypes:
id
analysistext_id
analysistype_id
示例数据：
entries
id year day
1 2010 1
2 2011 4
3 2012 6
AnalysisText
id text entries_id
1 blah 2
2 more blah 1
3 blah blah 1
AnalysisType
id type
1 A
2 B
3 C
AnalysisTextTypes
id analysistext_id analysistype_id
1 1 1
2 1 2
3 2 3
4 3 1
我需要获取具有多个analysistext的所有analysistype值，以及与这些entries相关联且具有多个analysistext的所有analysistype my_date text type
2011-4 blah A
2011-4 blah B
analysisText 1}}。
所以在上面的数据中，我们想要结果
analysisType因为那是具有多种类型的文本。
下面的查询有效，但使用子查询来识别超过select
CONCAT(e.year, '-', e.day) as my_date,
x.text,
y.type
from entries e
join analysistext x on e.id = x.entries_id
join analysistext_types xy on xy.analysistext_id = x.id
join analysistype y on y.id = xy.analysistype_id
where xy.analysistext_id in (
select analysistext_id
from analysistext_types
group by analysistext_id
having count(analysistype_id) > 1
) order by my_date, analysis_text, type;
的{{1}}值感觉效率低下：
SELECT
我想要一个查询，其中我们group by来自连接表的感兴趣字段，并直接对要连接的表执行having count和pandas。这可能吗？
答案 0 :(得分：0)
您可以使用自联接来完全避免子查询
select concat(e.year, '-' ,e.day), at.text ,ant.type
from AnalysisText at
join entries e on at.entries_id = e.id
join AnalysisTextTypes att on att.analysistext_id = at.id
join AnalysisType ant on att.analysistype_id = ant.id
join AnalysisTextTypes att2 on att2.analysistext_id = att.analysistext_id
group by att2.analysistext_id
having count(att2.analysistext_id) > 1