美文网首页
Mysql 分组查询top n(多种方法)

Mysql 分组查询top n(多种方法)

作者: 刘烟火 | 来源:发表于2020-09-08 12:00 被阅读0次

    情况说明

    查询每门科目的前3名。

    对于查询的结果,一般有两种情况。

    表所有数据为:

    image-20200830171608567

    情况1:对于分数相同的人,其后面的人 紧跟着名次排,直到排够名次3,就不再往后取了。

    image-20200830170431328

    情况2:对于分数相同的人,若当前相同名次的人数大于或等于 3, 则相同分数其后面的人不再参于top3了。

    image-20200830170939339

    在情况2中 ,为什么”李四 - java“ 这行没有了呢?

    可以这样理解,在情况2中相当于使用了名次空缺, 分数相同的人其后面那个人,的名次为 前面的人数+1 , 这里的 ”李四-java“这行,他的名次应当是 5 , 所有top3自然取不到 ”李四-java“这行。

    # 查询方法

    情况1方法:

    # 方法1
    SELECT      T1.*
    FROM        score_t T1
    WHERE       (SELECT COUNT(distinct score) FROM score_t WHERE T1.subject = subject AND T1.score < score) < 3
    ORDER BY    subject,score DESC;
    -- 思路:
    -- 1. 通过子查询的方式 , 查询每条记录中 对应科目比他分数高的 
    -- 2. 如果比自己还高的 小于3 ,那么自己就是属于top3
    
    
    
    # 方法2
     SELECT     T1.*
    FROM        score_t T1
    LEFT JOIN   (SELECT DISTINCT subject, score FROM score_t) T2
    ON          T1.subject = T2.subject
    AND         T1.score < T2.score
    GROUP BY    name,subject,score
    HAVING      COUNT(1) < 3
    ORDER BY    subject,score DESC;
    
    -- 思路:
    -- 1. 找到所有科目对应分数的所有情况 (去重)
    -- 2. 和原表关联后, 且比当前 记录 对应科目比自己分数
    -- 3. 分组后 找出 次数小于3的
    -- 注意: 使用 distinct 去重 和不使用, 区别于并列名次 是否算同一个名额
    
    # 方法3
    select t.name,t.subject,t.score,t.rank from (
        select name,subject,score,
        @r:= if(@psub = subject, if(@psco = score, @r, @r + 1), 1) rank ,
        @psub:= subject , @psco:= score 
        from score_t, (select @psub:=null, @psco:=null, @r:=0) init
        order by subject,score desc
    ) t where t.rank <=3;
    -- 思路:
    -- 使用 临时变量, 相同的分数 保持名次不变,否则+1
    
    
    
    # 方法4
    select t1.name,t1.subject,t1.score, count(distinct t2.score)  rank from score_t t1 -- 
    left join score_t t2
    on t1.subject = t2.subject and t2.score >= t1.score
    group by t1.subject,t1.name,t1.score
    having count(distinct t2.score) <= 3
    order by subject,score desc ,rank;
    -- 和方法2思路一样 , 只是使用 写法不同
    

    情况2方法:

    # 方法1:  
    SELECT      T1.*
    FROM        score_t T1
    WHERE       (SELECT COUNT(1) FROM score_t WHERE T1.subject = subject AND T1.score < score) < 3
    ORDER BY    subject,score DESC;
    
     
    
    # 方法2    
     SELECT     T1.*
    FROM        score_t T1
    LEFT JOIN   (SELECT subject, score FROM score_t) T2
    ON          T1.subject = T2.subject
    AND         T1.score < T2.score
    
    GROUP BY    name,subject,score
    HAVING      COUNT(1) < 3
    ORDER BY    subject,score DESC
    
    
    # 方法3
    select t.name,t.subject,t.score,t.rank from (
        select name,subject,score,
        @allorder:= if(@psub = subject, @allorder + 1, 1),
        @r:= if(@psub = subject, if(@psco = score, @r, @allorder), 1) rank ,    
        @psub:= subject , @psco:= score
        from score_t, (select @psub:=null, @psco:=null, @r:=0, @allorder:=0) init
        order by subject,score desc
    ) t where t.rank <=3
    
    
    # 方法4
    select t1.name,t1.subject,t1.score  from score_t t1 -- 
    left join score_t t2
    on t1.subject = t2.subject and t2.score > t1.score
    group by t1.subject,t1.name,t1.score
    having count(*) <  3
    order by subject,score desc 
    
    

    情况2和情况 写法都是对应的, 需要注意的是 并列名次,后面的人 是否需要保持名次空缺。那么统计的时候就需要根据情况去重。

    数据创建

    CREATE TABLE score_t(
    name        VARCHAR(10),
    subject     VARCHAR(10),
    score       INT(10)
    );
    
    INSERT INTO score_t VALUES
    ("张三","java",71),
    ("李四","java",82),
    ("王五","java",90),
    ("赵六","java",98),
    ("孙七","java",90),
    ("陈八","java",90),
    ("张三","python",81),
    ("李四","python",92),
    ("王五","python",93),
    ("赵六","python",97),
    ("刘九","python",93),
    ("孙七","python",70);
    

    参考链接:

    https://blog.csdn.net/weixin_44497013/article/details/107317719

    https://leetcode-cn.com/problems/department-top-three-salaries/solution/mysqlzi-ding-yi-bian-liang-shi-xian-fen-lei-pai-xu/

    相关文章

      网友评论

          本文标题:Mysql 分组查询top n(多种方法)

          本文链接:https://www.haomeiwen.com/subject/auixektx.html