sql窗口函数（详细版本）

mysql在8.0之后版本支持窗口函数。

使用的数据集为python中自带的红酒数据集。红酒数据集包括13个特征，178个样本，最后将红酒分为三类。

#导入数据
USE mydb;
DROP TABLE IF EXISTS wine
CREATE TABLE wine(alcohol FLOAT, malic_acid FLOAT, ash FLOAT, alcalinity_of_ash FLOAT, magnesium INT,
       total_phenols FLOAT, flavanoids FLOAT, nonflavanoid_phenols FLOAT,
       proanthocyanins FLOAT, color FLOAT, hue FLOAT, od280 FLOAT, proline INT, target INT)
       
LOAD DATA LOCAL INFILE "C://wine.csv" INTO TABLE wine
 FIELDS TERMINATED BY ',' OPTIONALLY ENCLOSED BY '"' ESCAPED BY '"'
LINES TERMINATED BY '\n';

窗口函数大体可以分为以下两种：

1.能够作为窗口函数的聚合函数（sum，avg，count，max，min）

2.rank，dense_rank。row_number等专用窗口函数

语法格式

<窗口函数> over ([partition by <列清单>]
                        order by <排序用列清单>)

一、sum

partition by 根据某些条件分开求和，order by 根据某条件是累加求和。

条件可以为一个也可以为多个。

#以taeget作为分类标准，将酒精度数求和
SELECT alcohol,target,
SUM(alcohol) over(PARTITION BY target)AS al_sum
FROM wine

SELECT alcohol,target,
SUM(alcohol) over(ORDER BY target)AS al_sum
FROM wine

SELECT alcohol,target,ash,
SUM(alcohol) over(PARTITION BY target,ash )AS al_sum
FROM wine

SELECT alcohol,target,ash,
SUM(alcohol) over(PARTITION BY target ORDER BY ash )AS al_sum
FROM wine

二、avg

partition by 分组后的均值，order by 是累加求和后计算得到的均值。

count，max，min同理。

SELECT alcohol,target,ash,
AVG(alcohol) over(PARTITION BY target,ash )AS al_avg
FROM wine

SELECT alcohol,target,ash,
AVG(alcohol) over(ORDER BY target,ash)AS al_avg
FROM wine

三、rank

1.rank函数：计算排序时，如果存在相同位次的记录，则会跳过之后的位次。

2.dense_rank函数：同样是计算排序，即使存在相同位次的记录，也不会跳过之后的位次。

3.row_number函数：赋予唯一的连续位次。
 

SELECT ash,
rank() over(ORDER BY ash)AS ash_rank,
dense_rank() over(ORDER BY ash)AS ash_dense_rank,
row_number() over (ORDER BY ash) AS ash_row_num
FROM wine

四、其他

unbounded：无界限
preceding：从分区第一行头开始，则为 unbounded。 N为：相对当前行向前的偏移量
following ：与preceding相反,到该分区结束，则为 unbounded。N为：相对当前行向后的偏移量
current row：顾名思义，当前行，偏移量为0
 

SELECT alcohol,ash,
COUNT(alcohol) over(PARTITION BY ash ROWS BETWEEN unbounded preceding AND current ROW)AS pv1,
COUNT(alcohol) over(PARTITION BY ash ROWS BETWEEN 3 preceding AND current ROW) AS pv2,
COUNT(alcohol) over(PARTITION BY ash ROWS BETWEEN 3 preceding AND 1 following) AS pv3,
COUNT(alcohol) over(PARTITION BY ash ROWS BETWEEN current ROW AND unbounded following) AS pv4

FROM wine

我把代码都写在上面了，如果有看不太懂的可以自己运行试一下，改改语法什么的，很容易就可以理解了。