我使用下表的解决方案在MySQL:

CREATE TABLE transactions (
  transaction_id int , user_id int , merchant_name varchar(255), transaction_date date , amount int
);

INSERT INTO transactions (transaction_id, user_id, merchant_name, transaction_date, amount)  
VALUES (1, 1 ,'abc', '2015-08-17', 100),(2, 2, 'ced', '2015-2-17', 100),(3, 1, 'def', '2015-2-16', 121),
(4, 1 ,'ced', '2015-3-17', 110),(5, 1, 'ced', '2015-3-17', 150),(6, 2 ,'abc', '2015-4-17', 130), 
(7, 3 ,'ced', '2015-12-17', 10),(8, 3 ,'abc', '2015-8-17', 100),(9, 2 ,'abc', '2015-12-17', 140),(10, 1,'abc', '2015-9-17', 100),
(11, 1 ,'abc', '2015-08-17', 121),(12, 2 ,'ced', '2015-12-23', 130),(13, 1 ,'def', '2015-12-23', 13),(3, 4, 'abc', '2015-2-16', 120),(3, 4, 'def', '2015-2-16', 121),(3, 4, 'ced', '2015-2-16', 121);

计算“金额”列的中位数:

WITH Numbered AS 
(
SELECT *, COUNT(*) OVER () AS TotatRecords,
    ROW_NUMBER() OVER (ORDER BY amount) AS RowNum
FROM transactions
)
SELECT Avg(amount)
FROM Numbered
WHERE RowNum IN ( FLOOR((TotatRecords+1)/2), FLOOR((TotatRecords+2)/2) )
;

TotalRecords = 16 and Median = 120.5000

此查询将适用于两种情况，即偶数和奇数记录。

ORACLE的简单解决方案:

SELECT ROUND(MEDIAN(Lat_N), 4) FROM Station;

简单的解决方案，理解MySQL:

select case MOD(count(lat_n),2) 
when 1 then (select round(S.LAT_N,4) from station S where (select count(Lat_N) from station where Lat_N < S.LAT_N ) = (select count(Lat_N) from station where Lat_N > S.LAT_N))
else (select round(AVG(S.LAT_N),4) from station S where 1 = (select count(Lat_N) from station where Lat_N < S.LAT_N ) - (select count(Lat_N) from station where Lat_N > S.LAT_N))
end from station;

解释

STATION是表名。LAT_N是具有数值的列名

假设站表中有101条记录(奇数)。这意味着如果表以asc或desc排序，则中位数是第51条记录。

In above query for every S.LAT_N of S table I am creating two tables. One for number of LAT_N values less than S.LAT_N and another for number of LAT_N values greater than S.LAT_N. Later I am comparing these two tables and if they are matched then I am selecting that S.LAT_N value. When I check for 51st records there are 50 values less than 51st record and there 50 records greater than 51st record. As you see, there are 50 records in both tables. So this is our answer. For every other record there are different number of records in two tables created for comparison. So, only 51st record meets the condition.

现在假设站表中有100条记录(偶数)。这意味着如果表以asc或desc排序，则中位数是第50条和第51条记录的平均值。

Same as odd logic I am creating two tables. One for number of LAT_N values less than S.LAT_N and another for number of LAT_N values greater than S.LAT_N. Later I am comparing these two tables and if their difference is equal to 1 then I am selecting that S.LAT_N value and find the average. When I check for 50th records there are 49 values less than 50th record and there 51 records greater than 50th record. As you see, there is difference of 1 record in both tables. So this(50th record) is our 1st record for average. Similarly, When I check for 51st records there are 50 values less than 51st record and there 49 records greater than 51st record. As you see, there is difference of 1 record in both tables. So this(51st record) is our 2nd record for average. For every other record there are different number of records in two tables created for comparison. So, only 50th and 51st records meet the condition.

如果这是MySQL，现在有窗口函数，你可以这样做(假设你想四舍五入到最接近的整数-否则只需将round替换为CEIL或FLOOR或其他什么)。下面的解决方案适用于表，无论表的行数是偶数还是奇数:

WITH CTE AS (
    SELECT val,
            ROW_NUMBER() OVER (ORDER BY val ASC) AS rn,
            COUNT(*) OVER () AS total_count
    FROM data
)
SELECT ROUND(AVG(val)) AS median
FROM CTE
WHERE
    rn BETWEEN
    total_count / 2.0 AND
    total_count / 2.0 + 1;

I think some of the more recent answers on this thread were already getting at this approach, but it also seemed like people were overthinking it, so consider this an improved version. Regardless of SQL flavor, there is no reason anyone should be writing a huge paragraph of code with multiple subqueries just to get the median in 2021. However, please note that the above query only works if you're asked to find the median for a continuous series. Of course, regardless of row number, sometimes people do make a distinction between what is referred to as the Discrete Median and what is referred to as the Interpolated Median for a continuous series.

如果你被要求为一个离散级数找到中位数，而表的行数是偶数，那么上面的解决方案就不适合你，你应该恢复使用其他解决方案之一，比如TheJacobTaylor的。

下面的第二个解决方案是对TheJacobTaylor的稍微修改的版本，其中我显式地声明了CROSS JOIN。这个方法也适用于行数为奇数的表，不管你是被要求求连续序列的中位数还是离散序列的中位数，但我特别会在被要求求离散序列的中位数时使用这个方法。否则，使用第一种解决方案。这样，您就永远不必考虑数据是包含“偶数”还是“奇数”个数的数据点。

SELECT x.val AS median
FROM data x
CROSS JOIN data y
GROUP BY x.val
HAVING SUM(SIGN(1 - SIGN(y.val - x.val))) = (COUNT(*) + 1) / 2;

最后，你可以在PostgreSQL中使用内置函数轻松做到这一点。这里有一个很好的解释，以及关于离散中位数和插值中位数的有效总结。

https://leafo.net/guides/postgresql-calculating-percentile.html#calculating-the-median

我刚刚在网上的评论中找到了另一个答案:

对于几乎所有SQL中的中位数: SELECT x.val from data x, data y GROUP BY x.val 总和(符号(1-SIGN (y.val-x.val))) = (COUNT (*) + 1) / 2

确保列有良好的索引，并且索引用于筛选和排序。与解释计划核对。

select count(*) from table --find the number of rows

计算“中值”行号。可能使用:median_row = floor(count / 2)。

然后把它从列表中挑出来:

select val from table order by val asc limit median_row,1

这将返回您想要的值的一行。

我有一个包含大约10亿行的数据库，我们需要它来确定集合中的年龄中位数。对十亿行进行排序是困难的，但如果你将可以找到的不同值(年龄范围从0到100)聚合在一起，你可以对这个列表进行排序，并使用一些算术魔术来找到你想要的任何百分位数，如下所示:

with rawData(count_value) as
(
    select p.YEAR_OF_BIRTH
        from dbo.PERSON p
),
overallStats (avg_value, stdev_value, min_value, max_value, total) as
(
  select avg(1.0 * count_value) as avg_value,
    stdev(count_value) as stdev_value,
    min(count_value) as min_value,
    max(count_value) as max_value,
    count(*) as total
  from rawData
),
aggData (count_value, total, accumulated) as
(
  select count_value, 
    count(*) as total, 
        SUM(count(*)) OVER (ORDER BY count_value ROWS UNBOUNDED PRECEDING) as accumulated
  FROM rawData
  group by count_value
)
select o.total as count_value,
  o.min_value,
    o.max_value,
    o.avg_value,
    o.stdev_value,
    MIN(case when d.accumulated >= .50 * o.total then count_value else o.max_value end) as median_value,
    MIN(case when d.accumulated >= .10 * o.total then count_value else o.max_value end) as p10_value,
    MIN(case when d.accumulated >= .25 * o.total then count_value else o.max_value end) as p25_value,
    MIN(case when d.accumulated >= .75 * o.total then count_value else o.max_value end) as p75_value,
    MIN(case when d.accumulated >= .90 * o.total then count_value else o.max_value end) as p90_value
from aggData d
cross apply overallStats o
GROUP BY o.total, o.min_value, o.max_value, o.avg_value, o.stdev_value
;

这个查询取决于你的db支持窗口函数(包括ROWS UNBOUNDED precede)，但如果你没有，这是一个简单的事情，将aggData CTE与自身连接，并将所有先前的总数聚合到' cumulative '列，用于确定哪个值包含指定的预分词。上面的示例计算p10、p25、p50(中位数)、p75和p90。

屁股的

我发现这个答案非常有用——https://www.eversql.com/how-to-calculate-median-value-in-mysql-using-a-simple-sql-query/

SET @rowindex := -1;

SELECT
   AVG(g.grade)
FROM
   (SELECT @rowindex:=@rowindex + 1 AS rowindex,
       grades.grade AS grade
    FROM grades
    ORDER BY grades.grade) AS g
WHERE
g.rowindex IN (FLOOR(@rowindex / 2) , CEIL(@rowindex / 2));