如何联合汇总查询？

我有一个查询如何联合汇总查询？

SELECT 
    CntApp = COUNT(app.ApplicationID) 
    ,r.RegionName 
    ,d.DistrictName 
FROM dim.Application app 
    JOIN dim.Geography g ON (app.ApplicationID = g.GeographyID) 
     AND (app.CountryId = g.CountryId) 
    JOIN dim.Region r ON r.RegionID = g.RegionID 
    JOIN dim.District d ON d.DistrictId = g.DistrictID 
    JOIN dim.ZIPcode z ON g.ZIPcodeID = z.ZIPcodeID   
GROUP BY 
    r.RegionName 
    ,d.DistrictName

和

SELECT 
    CntCon = COUNT(c.ContractID) 
    ,r.RegionName 
    ,d.DistrictName 
FROM dim.Contract c 
    JOIN dim.Geography g ON (c.ContractID = g.GeographyID) 
     AND (c.CountryId = g.CountryId) 
    JOIN dim.Region r ON r.RegionID = g.RegionID 
    JOIN dim.District d ON d.DistrictId = g.DistrictID 
    JOIN dim.ZIPcode z ON g.ZIPcodeID = z.ZIPcodeID   
GROUP BY 
    r.RegionName 
    ,d.DistrictName

，我要合并到一个表，所以在GROUP BY仍然有效。结果我想：

CntApp | CntCon | RegionName | DistrictName 
31  24  Pardubicky Pardubice 
21  16  Pardubicky Chrudim 
...

我试过UNION ALL却得到了这样的事情，而不是：

CntApp | CntCon | RegionName | DistrictName 
    NULL  24  Pardubicky Pardubice 
    21  NULL  Pardubicky Pardubice 
    26  NULL  Pardubicky Chrudim 
    ...

Appereantly，StackOverflow上认为我应该包括更多的文字（“它看起来像你的岗位主要是代码，请添加一些更多的细节“），所以在这里它是：

,-""-. 
    :======: 
    :======; 
    `-.,-' 
     || 
    _,''--. _____ 
    (/ __ `._| 
    ((_/_)\  | 
    (____)`.___| 
    (___)____.|_____ 
       SSt

来源

2017-07-03 Stanislav Jirák

所以，你得到了什么呢，当你UNION这一切？（好的绘画+1和幽默感;） –

@Rafal：看编辑。 –

当查询几乎完全相同时，您可以通过不进行UNION来优化您的查询;你有效地要求数据库加入同一组表并计算两次。有办法避免这种情况...（我发布了一个无联合答案作为替代） –

UNION ALL将列与结果相结合柱。您需要引入假列，再次聚集它（或其他溶液加入等）：

SELECT SUM(CntApp) CntApp, SUM(CntCon) CntCon, RegionName, DistrictName FROM (

    SELECT 
     CntApp = COUNT(app.ApplicationID) 
     ,CntCon = 0 
     ,r.RegionName 
     ,d.DistrictName 
    FROM dim.Application app 
     JOIN dim.Geography g ON (app.ApplicationID = g.GeographyID) 
      AND (app.CountryId = g.CountryId) 
     JOIN dim.Region r ON r.RegionID = g.RegionID 
     JOIN dim.District d ON d.DistrictId = g.DistrictID 
     JOIN dim.ZIPcode z ON g.ZIPcodeID = z.ZIPcodeID   
    GROUP BY 
     r.RegionName 
     ,d.DistrictName 

    UNION ALL 

    SELECT 
     CntApp = 0 
     ,CntCon = COUNT(c.ContractID) 
     ,r.RegionName 
     ,d.DistrictName 
    FROM dim.Contract c 
     JOIN dim.Geography g ON (c.ContractID = g.GeographyID) 
      AND (c.CountryId = g.CountryId) 
     JOIN dim.Region r ON r.RegionID = g.RegionID 
     JOIN dim.District d ON d.DistrictId = g.DistrictID 
     JOIN dim.ZIPcode z ON g.ZIPcodeID = z.ZIPcodeID   
    GROUP BY 
     r.RegionName 
     ,d.DistrictName 
) d 
GROUP BY RegionName, DistrictName

来源

2017-07-03 09:02:13

我有同样的想法，除了我用NULL而不是CntApp = 0。不幸的是，它为CntApp和CntCont返回相同的值，不应该这样做。 –

如果这没有回答你的问题，为什么你将它标记为答案？ –

@StanislavJirák - 你的意思是我提供的查询返回相同的值，或者一个是你使用了空值？ –

您需要join 2子查询。这样，您将按照您的期望并排查询两个查询的列。

这应该工作：

SELECT iq1.CntApp , iq2.CntCon, iq1.iq1.RegionName,iq1.DistrictName 
FROM 
(
SELECT 
    CntApp = COUNT(app.ApplicationID) 
    ,r.RegionName 
    ,d.DistrictName 
FROM dim.Application app 
    JOIN dim.Geography g ON (app.ApplicationID = g.GeographyID) 
     AND (app.CountryId = g.CountryId) 
    JOIN dim.Region r ON r.RegionID = g.RegionID 
    JOIN dim.District d ON d.DistrictId = g.DistrictID 
    JOIN dim.ZIPcode z ON g.ZIPcodeID = z.ZIPcodeID   
GROUP BY 
    r.RegionName 
    ,d.DistrictName 
) iq1 
inner join 
(
SELECT 
    CntCon = COUNT(c.ContractID) 
    ,r.RegionName 
    ,d.DistrictName 
FROM dim.Contract c 
    JOIN dim.Geography g ON (c.ContractID = g.GeographyID) 
     AND (c.CountryId = g.CountryId) 
    JOIN dim.Region r ON r.RegionID = g.RegionID 
    JOIN dim.District d ON d.DistrictId = g.DistrictID 
    JOIN dim.ZIPcode z ON g.ZIPcodeID = z.ZIPcodeID   
GROUP BY 
    r.RegionName 
    ,d.DistrictName 
) iq2 
on 
iq1.RegionName = iq2.iq1.RegionName 
and 
iq1.DistrictName = iq2.DistrictName

来源

2017-07-03 09:01:11

INNER JOIN是一个明智的选择吗？有一种风险是，一个子查询将不会有特定区域/区域配对的行，因此行将从输出中丢失 –

这就是为什么我已经突出显示顶部的“加入”:)直到我们没有从OP得到适当的数据，我们不能评论左/右或内心。我选择inner作为示例，因为它不会返回任何NULL列。 –

你需要一个FULL JOIN

SELECT coalesce(app.RegionName, c.RegionName) AS RegionName, 
     coalesce(app.DistrictName, c.DistrictName) AS DistrictName, 
     coalesce(app.CntApp,0) AS CntApp, 
     coalesce(c.CntCon,0) AS CntCon 
FROM 
    (SELECT 
     CntApp = COUNT(app.ApplicationID) 
     ,r.RegionName 
     ,d.DistrictName 
    FROM dim.Application app 
     JOIN dim.Geography g ON (app.ApplicationID = g.GeographyID) 
     AND (app.CountryId = g.CountryId) 
     JOIN dim.Region r ON r.RegionID = g.RegionID 
     JOIN dim.District d ON d.DistrictId = g.DistrictID 
     JOIN dim.ZIPcode z ON g.ZIPcodeID = z.ZIPcodeID   
    GROUP BY 
     r.RegionName 
     ,d.DistrictName 
    ) app 
    FULL JOIN 
    (
     SELECT 
     CntCon = COUNT(c.ContractID) 
     ,r.RegionName 
     ,d.DistrictName 
    FROM dim.Contract c 
     JOIN dim.Geography g ON (c.ContractID = g.GeographyID) 
     AND (c.CountryId = g.CountryId) 
     JOIN dim.Region r ON r.RegionID = g.RegionID 
     JOIN dim.District d ON d.DistrictId = g.DistrictID 
     JOIN dim.ZIPcode z ON g.ZIPcodeID = z.ZIPcodeID   
    GROUP BY 
     r.RegionName 
     ,d.DistrictName 
    ) c ON app.RegionName = c.RegionName AND app.DistrictName = c.DistrictName

来源

2017-07-03 09:05:23 TriV

你也许能够摆脱工会，这将是更安全因为你的结果不会被流浪笛卡尔的影响加入，如果坏数据打交道的方式进入G/R/d/Z表可能会发生：

SELECT 
CntApp, 
CntCon, 
r.RegionName, 
d.DistrictName 
FROM 
dim.Geography g 
INNER JOIN dim.Region r ON r.RegionID = g.RegionID 
INNER JOIN dim.District d ON d.DistrictId = g.DistrictID 
INNER JOIN dim.ZIPcode z ON g.ZIPcodeID = z.ZIPcodeID 

LEFT JOIN (SELECT ApplicationID, CountryID, COUNT(*) CntApp FROM dim.Application GROUP BY ApplicationID, CountryID) app 
    ON (app.ApplicationID = g.GeographyID) AND (app.CountryId = g.CountryId) 

LEFT JOIN (SELECT ContractID, CountryId, COUNT(*) as CntCon FROM dim.Contract GROUP BY ContractID, CountryId) c  
    ON (c.ContractID = g.GeographyID) AND (c.CountryId = g.CountryId)

这里有点教育点为你虽然：

如果你有两个数据块（来自表格，查询等等），并且你想把它们垂直放在一起（更多的行），那么你使用UNION 如果你想水平地把它们合并（更多列）使用JOIN

如果我们有：

a,b,c 
a,b,c 
a,b,c

而且

a,y,z 
a,y,z 
a,y,z

这是你与UNION得到：

a,b,c 
a,b,c 
a,b,c 
a,y,z 
a,y,z 
a,y,z

这是你会得到什么JOIN：

a,b,c,y,z 
a,b,c,y,z 
a,b,c,y,z

记住这一点，是将竭诚为您服务以及

来源

2017-07-03 09:31:43

如何联合汇总查询？

回答

相关问题