SQL DENSE_RANK 和 PARTITION BY

Question

我不完全理解下面列出的查询结果：

CREATE TABLE #tmpAccountsTable (ID INT PRIMARY KEY, AccountID INT, 

AccountName varchar(20), CompanyID int, CompanyName varchar(50))
INSERT INTO #tmpAccountsTable (ID, AccountID, AccountName, CompanyID, CompanyName)
SELECT 1 AS ID, 1 AS AccountID, 'ABC Account' AS AccountName, 1 AS CompanyID, 'Stark Enterprise' AS CompanyName UNION
SELECT 2 AS ID, 1 AS AccountID, 'ABC Account' AS AccountName, 1 AS CompanyID, 'Stark Enterprise' AS CompanyName UNION
SELECT 3 AS ID, 2 AS AccountID, 'DEF Account' AS AccountName, 1 AS CompanyID, 'Stark Enterprise' AS CompanyName UNION
SELECT 4 AS ID, 3 AS AccountID, 'GHI Account' AS AccountName, 1 AS CompanyID, 'Stark Enterprise' AS CompanyName UNION
SELECT 5 AS ID, 4 AS AccountID, 'JKL Account' AS AccountName, 1 AS CompanyID, 'Stark Enterprise' AS CompanyName UNION
SELECT 6 AS ID, 1 AS AccountID, 'ABC Account' AS AccountName, 1 AS CompanyID, 'Stark Enterprise' AS CompanyName UNION
SELECT 7 AS ID, 1 AS AccountID, 'ABC Account' AS AccountName, 1 AS CompanyID, 'Stark Enterprise' AS CompanyName UNION
SELECT 8 AS ID, 1 AS AccountID, 'ABC Account' AS AccountName, 1 AS CompanyID, 'Stark Enterprise' AS CompanyName UNION
SELECT 9 AS ID, 2 AS AccountID, 'DEF Account' AS AccountName, 1 AS CompanyID, 'Stark Enterprise' AS CompanyName

SELECT DENSE_RANK() OVER (PARTITION BY AccountID ORDER BY CompanyName) as AccountRANK, * FROM #tmpAccountsTable

DROP TABLE #tmpAccountsTable

执行上述，我得到以下结果集：

根据我对SELECT DENSE_RANK() OVER (PARTITION BY AccountID ORDER BY CompanyName) as AccountRANK, * FROM #tmpAccountsTable 的理解，应该为所有相同的AccountId 创建一个分区，并且它们应该收到相同的DENSE_RANK() 值，但可以看出，情况并非如此。

我错过了什么？

Answer 1

如果您希望每个 AccountId 有一个单独的值，那么它应该是 ORDER BY 的一部分，而不是 PARTITION BY：

SELECT DENSE_RANK() OVER (ORDER BY AccountID, CompanyName) as AccountRANK, *
FROM #tmpAccountsTable;

PARTITION BY 用于重置计数器，因此每个AccountId从 1 开始。然后将CompanyNames 排序。因为它们都有相同的值，所以结果是1。