postgresql - 来自别名的 SQL 累计和

Question

我想根据我在同一个查询中生成的百分比计算总和。我知道做累积的方法:

select sum(grandtotal)over(order by agentname) as cumulative from data

但是现在我想做累积的列还没有在数据库中。它是在同一个查询中生成的(别名:百分比)

 SELECT 
 agentname,weakness,
 count(*) as frequency,
 count(*)*100.00/sum(count(*))over(order by agentname) as percentage
 from ... where ...

然后我尝试:

(select sum(percentage)over(order by agentname) as cumulative from data

错误提示“百分比”列不存在。我如何应用累计金额？谢谢

这是我想要的输出的表格:

    agentname | weakness | frequency | percentage | cumulative
       A      |   W1     |     4     |    36.36   |    36.36
       A      |   W2     |     4     |    36.36   |    72.72
       A      |   W3     |     2     |    18.18   |    90.09
       A      |   W4     |     1     |     9.09   |     100

Answer 1

不能根据同一 SELECT(在大多数数据库中)中另一个窗口函数的结果来计算窗口函数。

您必须再次嵌套该查询:

SELECT t.*, SUM(percentage) OVER (ORDER BY agentname) AS cumulative
FROM (
  SELECT 
    agentname,
    weakness,
    COUNT(*) AS frequency,

    -- No ORDER BY in this SUM()!
    COUNT(*) * 100.00 / SUM(COUNT(*)) OVER () AS percentage
  FROM ... WHERE ...
) AS t
ORDER BY agentname

进一步反馈:

当你这样做时，我建议通过向它们添加另一列来使 ORDER BY 子句具有确定性，例如弱点。

此外，我不确定您的要求，但我认为这些百分比需要根据 agentname 计算？在这种情况下，您必须在 SUM(COUNT(*)) OVER(...) 窗口函数中添加 PARTITION BY agentname 子句。