Question

我有一个带有PlayerID和TypeID的棒球比赛数据库（比赛类型：双击，击出等）。数据看起来像这样：

+----------+--------+
| playerid | typeid |
+----------+--------+
|        2 |      4 |
|        2 |      4 |
|        2 |      7 |
|        3 |      7 |
|        3 |      7 |
|        3 |      7 |
|        3 |     26 |
|        3 |      7 |

我正在努力寻找哪种球员最能发挥各种类型。例如。 Jim（PlayerID 3）获得最多击球（TypeID 7）和Bob（PlayerID 2）拥有最多本垒打（TypeID 4），这应该产生下表：

+----------+--------+----------------+
| playerid | typeid | max(playcount) |
+----------+--------+----------------+
|        2 |      4 |             12 |
|        3 |      7 |              9 |
|        3 |     26 |              1 |

到目前为止，我最好的尝试是：

SELECT playerid,typeid,MAX(playcount) FROM 
(
SELECT playerid,typeid,COUNT(*) playcount FROM plays GROUP BY playerid,typeid
) AS t GROUP BY typeid;

返回每种类型的正确最大值，但关联的PlayerID都是错误的，我无法弄清楚原因。我敢肯定我错过了一些简单的东西（或者让它过于复杂），但无法弄明白。有什么想法吗？

Answer 1

在MySQL中，这个群体=明智的最大值，遗憾的是，它并不像你想要的那样简单。

以下是使用类似于ROW_NUMBER() in MySQL

中建议的方法来实现此目的的方法

SELECT a.*
  FROM (
        SELECT playerid
              ,typeid
              ,COUNT(*) playcount 
          FROM plays 
         GROUP BY playerid,typeid
        ) a
  LEFT JOIN
        (
        SELECT playerid
              ,typeid
              ,COUNT(*) playcount 
          FROM plays 
         GROUP BY playerid,typeid
        ) b
    ON a.typeid = b.typeid
   AND a.playcount < b.playcount
WHERE b.playerid IS NULL

Answer 2

这会有用吗？

SELECT
    playertypecounts.*
FROM
    (SELECT
       playerid,
       typeid,
       COUNT(*) as playcount
     FROM plays
     GROUP BY playerid, typeid) playertypecounts
INNER JOIN
    (SELECT
        typeid,
        MAX(playcount) as maxplaycount
     FROM 
        (SELECT
           playerid,
           typeid,
           COUNT(*) as playcount
         FROM plays
         GROUP BY playerid, typeid) playcounts
     GROUP BY typeid) maxplaycounts
ON playertypecounts.typeid = maxplaycounts.typeid
AND playertypecounts.playcount = maxplaycounts.maxplaycount

查询块的这一部分返回每个typeid的最大游戏次数：

(SELECT
    typeid,
    MAX(playcount) as maxplaycount
 FROM 
    (SELECT
       playerid,
       typeid,
       COUNT(*) as playcount
     FROM plays
     GROUP BY playerid, typeid) playcounts
 GROUP BY typeid) maxplaycounts

然后它内部连接到所有typeid / playcounts，以便过滤那些玩家对任何给定typeid具有最大计数的计数。

请参阅SQLFiddle example。

说了这么多，我实际上更喜欢@KarlKieninger的回答，因为它更优雅。

Answer 3

你必须将playerid列也放在group by子句中。休息一切都好。

SELECT playerid,typeid,MAX(playcount) FROM 
(
SELECT playerid,typeid,COUNT(*) playcount FROM plays GROUP BY playerid,typeid
) AS t GROUP BY playerid,typeid;

选择具有另一个ID

3 个答案: