运行 每条记录占总和的总百分比。 Presto/Athena/SQL

Running total percentage per record over total sum. Presto/Athena/SQL

我正在尝试计算每一行的累计百分比 Presto/Athena。例如:如果我有这样的数据

AccountID | UserID | HolidaysTaken
ABC       | A      | 4
ABC       | B      | 6
ABC       | B      | 3
ABC       | K      | 2
ABC       | K      | 3
ABC       | X      | 1

现在在 运行 这个查询之后我得到以下结果。

SELECT AccountID, UserID, sum(HolidaysTaken) AS HolidaysTaken FROM table
WHERE AccountID = 'ABC'
GROUP BY AccountID, UserID
ORDER BY HolidaysTaken DESC

AccountID | UserID | HolidaysTaken 
ABC       | B      | 9             
ABC       | K      | 5             
ABC       | A      | 4             
ABC       | X      | 1 

Total holiday taken by all users = 19         

但我想再添加 2 列。 EachUserPercentage:每个用户的假期占总假期的百分比。 CumulativePercentage:EachUserPercentage 的累计和。这个我可以用

AccountID | UserID | HolidaysTaken | EachUserPercentage | CumulativePercentage
ABC       | B      | 9             | 47.36              | 47.36  
ABC       | K      | 5             | 26.31              | 73.67
ABC       | A      | 4             | 21.05              | 94.72
ABC       | X      | 1             | 5.26               | 100

我尝试了 diff window 函数 percent_rank(), cume_dist() and ntile() 但无法正常工作 EachUserPercentage

您可以使用 window 函数查找 AccountID 的百分比,然后使用另一个 window 函数对按每个 UserID 休假总数排序的未绑定行求和。如下所示:

WITH totalUser
AS (SELECT   AccountID
            ,UserID
            ,SUM(HolidaysTaken) AS HolidaysTaken
            ,CAST(100.0 * SUM(HolidaysTaken) / SUM(SUM(HolidaysTaken)) OVER (PARTITION BY AccountID) AS NUMERIC(5, 2)) AS EachUserPercentage
    FROM     table
    WHERE    AccountID = 'ABC'
    GROUP BY AccountID
            ,UserID)
SELECT   totalUser.AccountID
        ,totalUser.UserID
        ,totalUser.HolidaysTaken
        ,totalUser.EachUserPercentage
        ,SUM(totalUser.EachUserPercentage) OVER (PARTITION BY totalUser.AccountID
                                                 ORDER BY totalUser.EachUserPercentage DESC
                                                 ROWS UNBOUNDED PRECEDING)
FROM     totalUser
ORDER BY totalUser.HolidaysTaken DESC;

您好,如果您的组在 AccountID(已考虑)上,您可以从下面的查询中简单地获取 EachUserPercentage。

SELECT table.AccountID, UserID, sum(table.HolidaysTaken) AS HolidaysTaken,
MAX(CAST(all_sum.HolidaysTaken AS NUMERIC(12,2))),
(SUM(CAST(table.HolidaysTaken AS NUMERIC(12,2)))/MAX(CAST(all_sum.HolidaysTaken AS NUMERIC(12,2))))*100 EachUserPercentage
 FROM table
LEFT OUTER JOIN (SELECT SUM(HolidaysTaken) AS HolidaysTaken,AccountID FROM table GROUP BY AccountID)all_sum ON all_sum.AccountID= table.AccountID
WHERE table.AccountID = 'ABC'
GROUP BY table.AccountID, UserID
ORDER BY HolidaysTaken DESC

它在我这边有效。