编写查找最常见名字（S）、错误的 SQL 查询

Question

给定一个名为 USERS_FAKE U 的表，我需要编写一个查询来查找最常见的名字。

我的问题是我只设法编写了一个只返回一个常见名字的查询。但是，如果名称“Bob”出现 6 次，使其成为最常见的名字，并且还有其他名称出现 6 次（如“Mike”），我的查询将只返回“Bob”。我的查询应该返回 Bob 和 Mike，因为它们的出现次数相同。

尝试 MCRE：

-- The table we are working with
CREATE TABLE USERS_FAKE (
    USER_ID NUMBER PRIMARY KEY,
    FIRST_NAME VARCHAR2(100) NOT NULL,
    LAST_NAME VARCHAR2(100) NOT NULL
);

-- Test Data, to test that we return the Most Common First Name(s)
-- (clearly there are 2 Most Common First Name(s): Bob and Mike)
INSERT INTO USERS_FAKE VALUES(5,  'Bob',  'Io');
INSERT INTO USERS_FAKE VALUES(6,  'Bob',  'Io');
INSERT INTO USERS_FAKE VALUES(7,  'Bob',  'Io');
INSERT INTO USERS_FAKE VALUES(8,  'Bob',  'Io');
INSERT INTO USERS_FAKE VALUES(9,  'Bob',  'Io');
INSERT INTO USERS_FAKE VALUES(10, 'Bob',  'Io');
INSERT INTO USERS_FAKE VALUES(11, 'Mike',  'Ganymede');
INSERT INTO USERS_FAKE VALUES(12, 'Mike',  'Ganymede');
INSERT INTO USERS_FAKE VALUES(13, 'Mike',  'Ganymede');
INSERT INTO USERS_FAKE VALUES(14, 'Mike',  'Ganymede');
INSERT INTO USERS_FAKE VALUES(15, 'Mike',  'Ganymede');
INSERT INTO USERS_FAKE VALUES(16, 'Mike',  'Ganymede');

我需要编写一个使用聚合 + WHERE 子句的简单查询，但您在 SQL 中将两者结合起来。这是在 Oracle 中，我必须使用 ROWNUM，LIMIT 在我的环境中不起作用。

Answer 1

WHERE 子句中不允许聚合。您希望查询返回与最常见名称的出现次数相关联的名称。相反，应该使用HAVING 子句，因为HAVING 子句用于在SQL 中聚合之后过滤组。

见下例：

SELECT DISTINCT FIRST_NAME, COUNT(FIRST_NAME) AS MOST_FREQUENT 
FROM USERS_FAKE
GROUP BY FIRST_NAME
HAVING COUNT(FIRST_NAME) = (SELECT COUNT(FIRST_NAME) AS NUM_FREQUENCY 
                            FROM USERS_FAKE
                            GROUP BY FIRST_NAME
                            ORDER BY NUM_FREQUENCY DESC
                            LIMIT 1)
;

另外我个人推荐row_number()：

SELECT T.FIRST_NAME
FROM (
       SELECT N.FIRST_NAME AS FIRST_NAME
            , ROW_NUMBER() OVER (ORDER BY N.NUM_FREQUENCY DESC) AS ROW_NUM
       FROM (
               SELECT FIRST_NAME
                    , COUNT(FIRST_NAME) AS NUM_FREQUENCY 
               FROM USERS_FAKE
               GROUP BY FIRST_NAME
               ORDER BY NUM_FREQUENCY
            ) AS N
      ) AS T
WHERE T.ROW_NUM = 1
;

ROW_NUMBER() 按出现次数对名称进行排名，然后查询返回与最常见名称相关的第一行。在这种情况下，ROW_NUMBER() 不是最佳解决方案。但是，这可能是实现“发生很重要”场景的一种简单方法。

Answer 2

SELECT n.first
FROM names n
JOIN ( SELECT first, COUNT(*) AS cnt
       FROM names
       GROUP BY first
     ) n1 ON ( n1.first = n.first )
ORDER BY n1.cnt DESC group by n.first;

Answer 3

我会使用 RANK() 函数

SELECT   t.first_name, t.count FROM
 (
   SELECT
   first_name, COUNT(1) AS count,
   RANK () OVER (  ORDER BY COUNT(1) DESC ) rank_no 
   FROM USERS_FAKE
   GROUP BY first_name
 ) AS t
 WHERE rank_no = 1