如何将查询中每个组的计数显示为百分比？

Question

我在 mytable(Movies) 中有一个专栏，演员。我正在尝试获取每个演员出现的次数百分比。

CREATE TABLE movies AS SELECT * FROM ( VALUES
  ('Robert DeSouza'),
  ('Tony Wagner'),
  ('Sean Cortese'),
  ('Robert DeSouza'),
  ('Robert DeSouza'),
  ('Tony Wagner'),
  ('Sean Cortese'),
  ('Charles Bastian'),
  ('Robert DeSouza')
) AS t(actors);

我要求的结果：

select Actors, (some formula * 100) as "The Ratio" from Movies

Actors                       The Ratio
Robert DeSouza                 44%
Tony Wagner                    22%
Sean Cortese                   22%
Charles Bastian                11%
                               100%

Answer 1

您可以使用window functions 执行此操作。对于数值计算：

select m.actor,
       count(*) * 1.0 / sum(count(*)) over () as ratio
from movies m
group by m.actor;

您可以将比率转换为您想要的任何格式——乘以 100 得到一个百分比，使用字符串连接来添加一个百分比。对我来说，所谓的比率应该介于 0 和 1 之间（在这种情况下）。

Answer 2

SELECT actors, floor(count(*) *100 / sum(count(*)) OVER ())
FROM movies
GROUP BY actors
ORDER BY count(*) DESC;

这可以让你大部分时间到达那里..

     actors      | floor 
-----------------+-------
 Robert DeSouza  |    44
 Tony Wagner     |    22
 Sean Cortese    |    22
 Charles Bastian |    11

不确定您的示例中如何获得 100。您想要百分比的下限，并且想要神奇地说 100？如果44+22+22+11 = 100 今天只是那些日子之一。但我们也可以这样做。

SELECT actors AS "Actors", r::text || '%' AS "The Ratio"
FROM (
  SELECT
    actors AS "Actors",
    floor(count(*) *100 / sum(count(*)) OVER ()) AS r,
    false AS is_total
  FROM movies
  GROUP BY actors
  UNION ALL
    SELECT *
    FROM ( VALUES
      (null, 100, true)
    ) AS t(actors, floor, is_total)
  ORDER BY 3, 2 DESC
) AS t(actors,r);

输出，

     Actors      | The Ratio 
-----------------+-----------
 Robert DeSouza  | 44%
 Tony Wagner     | 22%
 Sean Cortese    | 22%
 Charles Bastian | 11%
                 | 100%

如果你不想floor你可以round()

Answer 3

没有包含百分号（% 字符）的 numeric type，因此您的问题不能仅通过计算数值的表达式来解决。除了计算那个值，你还需要format it as text using the to_char() function。

此函数接受一个数值并使用您作为第二个参数提供的格式文字将其转换为文本值。在这种情况下，看起来您想要做的是四舍五入到最接近的百分比并显示百分号。您可能希望使用 '990%' 作为格式化文字。将此添加到您的示例表和the window function that Gordon suggested 产量：

[local] air@postgres=> CREATE TABLE movies AS SELECT * FROM ( VALUES
...   ('Robert DeSouza'),
...   ('Tony Wagner'),
...   ('Sean Cortese'),
...   ('Robert DeSouza'),
...   ('Robert DeSouza'),
...   ('Tony Wagner'),
...   ('Sean Cortese'),
...   ('Charles Bastian'),
...   ('Robert DeSouza')
... ) AS t(actors);

SELECT 9
Time: 715.613 ms
[local] air@postgres=> select actors, to_char(100 * count(*) / sum(count(*)) over (), '990%') as "The Ratio" from movies group by actors;
┌─────────────────┬───────────┐
│     actors      │ The Ratio │
├─────────────────┼───────────┤
│ Charles Bastian │   11%     │
│ Tony Wagner     │   22%     │
│ Sean Cortese    │   22%     │
│ Robert DeSouza  │   44%     │
└─────────────────┴───────────┘

(4 rows)

Time: 31.501 ms

您要确保考虑到显示所有可能值的需要，包括 100% 和 0%；由于to_char() 将四舍五入以适合您所需的精度，因此演员可以将其比率显示为零，尽管表中存在：

[local] air@postgres=> delete from movies where actors <> 'Tony Wagner';
DELETE 7
Time: 36.697 ms
[local] ahuth@postgres=> insert into movies (actors) select 'Not Tony Wagner' from generate_series(1,500);
INSERT 0 500
Time: 149.022 ms
[local] ahuth@postgres=> select actors, to_char(100 * count(*) / sum(count(*)) over (), '990%') as "The Ratio" from movies group by actors;
┌─────────────────┬───────────┐
│     actors      │ The Ratio │
├─────────────────┼───────────┤
│ Tony Wagner     │    0%     │
│ Not Tony Wagner │  100%     │
└─────────────────┴───────────┘
(2 rows)

Time: 0.776 ms

如果你想扩展它以显示小数位，只需修改格式字符串。当您想强制使用前导零或尾随零时，请在格式文字中使用 0。

Answer 4

这是通过使用联合将分组结果与整体结果连接起来来实现的。

编辑：在来自@Air 的 cmets 之后删除了串联和 0。

select actors actor, ratio from ( 
    select 0 sort
    , actors
    , round(count(*) * 100.0 / ( select count(*) 
    from movies ),0) ratio 
    from movies 
    group by actors 
    union 
    select 1 sort
    , 'Total'
    , round(count(*) / count(*) * 100,0) ratio 
    from movies 
) actors 
order by sort, actors ; 

--结果

      actor      | ratio 
-----------------+-------
 Charles Bastian | 11
 Robert DeSouza  | 44
 Sean Cortese    | 22
 Tony Wagner     | 22
 Total           | 100
(5 rows)