获取每组的第一行 SQL

Question

我有如下数据。我想从以下数据中得到每个性别的首选

subjectID <- c("1", "2", "1", "0", "1", "0", "0", "1", "0", "2", 
       "0", "0", "2", "2","2","1","2","1","0","2")
gender <- c("M", "M", "F", "M", "M", "F", "M", "M", "M", "F", 
    "M", "F", "M", "M", "F","M", "F", "M", "F",  "F")

 selection <- data.frame(subjectID, gender)
       subjectID <- c("1", "2", "0")
       subject <- c("Maths", "Music",  "English")

    subjects <- data.frame(subjectID, subject)

我试过按如下降序排列选项：

 favourite <- sqldf("SELECT  a.gender, b.subject, COUNT(a.subjectID) as `no of selections`
              FROM selection a
              JOIN subjects b
              ON (a.subjectID  = b.subjectID )
              GROUP BY a.subjectID, a.gender
              ORDER BY a.gender, `no of selections` DESC
              ")

但是，我想获得以下 table，其中我获得了每个性别的首选：

gender <- c("F", "M")
subjects <- c("Music", "Maths")
mostfav <- data.frame(gender, subjects)

Answer 1

如果我没理解错的话，你可以在SQL中使用window函数：

SELECT gs.*
FROM (SELECT s.gender, su.subject, COUNT(*) as cnt,
             ROW_NUMBER() OVER (PARTITION BY s.gender ORDER BY COUNT(*) DESC) as seqnum
      FROM selection s JOIN
           subjects su
           ON su.subjectID  = s.subjectID
      GROUP BY s.gender, su.subject
     ) gs
WHERE seqnum = 1;

Answer 2

如果您运行的是 MySQL 8.0，您可以在子查询中使用 RANK() 按每个性别的主题计数对记录进行排名，并在外部查询中过滤每组的顶部记录（如果有顶级关系，RANK() 保留它们）：

SELECT gender, subject, no_of_selections
FROM (
    SELECT  
        se.gender, 
        su.subject, 
        COUNT(*) as no_of_selections,
        RANK() OVER(PARTITION BY se.gender ORDER BY COUNT(*) DESC) rn
    FROM selection se
    JOIN subjects su ON se.subjectID  = su.subjectID
    GROUP BY se.subjectID, se.gender, su.subject
) t
WHERE rn = 1
ORDER BY gender DESC

在早期版本中，window 函数不可用，一种选择是使用 having 子句进行过滤，使 returns 每个性别的计数最高：

SELECT  
    se.gender, 
    su.subject, 
    COUNT(*) as no_of_selections
FROM selection se
JOIN subjects su ON se.subjectID  = su.subjectID
GROUP BY se.subjectID, se.gender, su.subject
HAVING COUNT(*) = (
    SELECT COUNT(*) 
    FROM selection se1
    WHERE se1.gender = se.gender
    GROUP BY se1.subjectID, se1.gender
    ORDER BY COUNT(*) DESC
    LIMIT 1
)

备注：

我更改了 table 别名，使它们更有意义
您应该将 subject 列添加到 GROUP BY 子句中，以使您的查询可以在 sql 模式 ONLY_FULL_GROUP_MODE 下运行，该模式默认启用开始 MySQL 5.7

获取每组的第一行 SQL

get first row fo each group SQL

mysql

sql

group-by

greatest-n-per-group

sqldf