1

我有一个 SQL 语句：

SELECT application.id,title,url,company.name AS company_name,package_name,ranking,date,platform,country.name AS country_name,collection.name AS collection_name,category.name AS category_name FROM application
JOIN application_history ON application_history.application_id = application.id
JOIN company ON application.company_id = company.id
JOIN country ON application_history.country_id = country.id
JOIN collection ON application_history.collection_id = collection.id
JOIN category ON application_history.category_id = category.id
WHERE application.platform=0
AND country.name ='CZ'
AND collection.name='topfreeapplications'
AND category.name='UTILITIES'
AND application_history.ranking <= 10
AND date::date BETWEEN date (CURRENT_DATE - INTERVAL '1 month') AND CURRENT_DATE
ORDER BY application_history.ranking ASC

它产生这个结果：

在此处输入图像描述

我想添加给定包的列平均排名和出现的列数，这将计算包出现在列表中的数量。我还想按 package_name 对结果进行分组，这样我就没有冗余了。

到目前为止，我尝试在 ORDER BY 之前添加一个 GROUP BY By 子句：

GROUP BY package_name

但它返回给我一个错误：

column "application.id" must appear in the GROUP BY clause or be used in an aggregate function

如果我添加它要求我添加的每一列，它就不起作用。我还尝试通过在 SELECT 之后添加来计算包名称的数量：

COUNT(package_name) AS count

它会产生类似的错误。

我怎样才能得到我正在寻找的结果？我应该改为进行两个查询，还是可以一次获取所有内容？我准确地说，我已经查看了关于 SO 的其他答案，但没有一个人试图在“生产”列上进行计数。

感谢您的帮助。

编辑 ：

这是我最初期望的结果：

在此处输入图像描述

虽然 Gordon 的建议没有给我正确的结果，但它让我走上了正轨，但当我读到这篇文章时：来自文档：“与常规聚合函数不同，使用窗口函数不会导致行分组为单个输出排。”

所以我又开始单独使用 COUNT 和 AVG。我的问题是我想显示排名列和日期以检查事情是否正确。但是，正如 Jarlh 在评论中提到的那样，将这些列放入 Select 会阻止 GROUP BY 按预期工作。

工作查询：

SELECT application.id,title,url,company.name AS company_name,package_name,platform,country.name AS country_name,collection.name AS collection_name,category.name AS category_name, 
    COUNT(package_name) AS count, AVG(application_history.ranking) AS avg
    FROM application
    JOIN application_history ON application_history.application_id = application.id
    JOIN company ON application.company_id = company.id
    JOIN country ON application_history.country_id = country.id
    JOIN collection ON application_history.collection_id = collection.id
    JOIN category ON application_history.category_id = category.id
    WHERE application.platform=0
    AND country.name ='CZ'
    AND collection.name='topfreeapplications'
    AND category.name='UTILITIES'
    AND application_history.ranking <= 10
    AND date::date BETWEEN date (CURRENT_DATE - INTERVAL '1 month') AND CURRENT_DATE
    GROUP BY package_name,application.id,company.name,country.name,collection.name,category.name
    ORDER BY count DESC
4

1 回答 1

1

我认为您需要窗口/分析功能。下面添加了两列，一列是每个包的行数，另一列是它们的平均排名：

SELECT application.id, title, url, company.name AS company_name, package_name, 
       ranking, date, platform, country.name AS country_name,
       collection.name AS collection_name, category.name AS category_name,
       count(*) over (partition by package_name) as count,
       avg(ranking) over (partition by package_name) as avg_package_ranking
FROM application . . .
于 2016-04-08T11:04:52.997 回答