使用排名计算各种平均值

Question

我的数据看起来像这样

+------+--------+------+-------+
| year | month  | name | value |
+------+--------+------+-------+
| 2017 |      1 | John |   100 |
| 2017 |      2 | Doe  |   200 |
| 2017 |      3 | Jane |   300 |
|    . |      . | .    |     . |
| 2018 |      1 | John |   150 |
| 2018 |      2 | Doe  |   250 |
| 2018 |      3 | Jane |   350 |
+------+--------+------+-------+

我正在尝试计算每年和每月的前 2 个名称的平均值。 我可以用下面的代码做到这一点

select year, month, avg(sum_value) as avg_of_2
from (
    select year,
             month,
             name,
             sum(value) as sum_value,
             rank() over (partition by year, month order by sum(value) desc) as rnk
      from database.table_a
      group by year, month, name
      order by 1, 2, 4 desc
    ) tbl_for_2
where rnk <= 2 -- for top 2 values
group by 1, 2
order by 1, 2;

但是现在我想将前 2 个名称的平均值扩展到前 5、10 和 50 个。有没有一种方法可以使用排名而不重复相同的查询？

我的最终结果就像

+------+-------+----------+----------+---------+
| year | month |  avg_2   |  avg_5   | avg_10  |
+------+-------+----------+----------+---------+
| 2017 |     1 | some_val | some_val | som_val |
| 2017 |     2 | some_val | some_val | som_val |
|   .. |       |          |          |         |
|   .. |       |          |          |         |
+------+-------+----------+----------+---------+

Answer 1

无法返回动态列数，必须写入所有列。 只需在外部查询中使用过滤聚合：

select year, month, 
  AVG(sum_value)FILTER(WHERE rnk<=2) as avg_2,
  AVG(sum_value)FILTER(WHERE rnk<=5) as avg_5,
  AVG(sum_value)FILTER(WHERE rnk<=10) as avg_10,
  .................
  AVG(sum_value)FILTER(WHERE rnk<=100) as avg_100,
  ... and so on
from (
    select year,
             month,
             name,
             sum(value) as sum_value,
             rank() over (partition by year, month order by sum(value) desc) as rnk
      from database.table_a
      group by year, month, name
    ) tbl
group by 1, 2
order by 1, 2;