如何对 google 云 sql 或 python 数据框中的每滚动 3 行进行分组？

Question

假设我有一个数据集。

Year | Month | Count
2017 | 1     | 3
2017 | 2     | 4
2017 | 3     | 2
2017 | 4     | 1
2017 | 5     | 5
2017 | 6     | 6

我希望按结果分组为

group 1    | total count
month1,2,3 |    9
month2,3,4 |    7
month3,4,5 |    8
...

我想知道在 google 云 SQL 中有没有办法做到这一点？如果没有，我可以在 python 数据框中执行此操作吗？

Answer 1

您可以使用 lead() 和 sum()。假设 month 是一个字符串：

select concat(month, ',', lead(month) over (order by year, month), ',',
              lead(month, 2) over (order by year, month)
             ),
       sum(count) over (order by year, month rows between current row and 2 following) as total
from t;

或：

with t as (
      select 2017 as year, 1 as month, 3 as count union all
      select 2017, 2, 4 union all
      select 2017, 3, 2 union all
      select 2017, 4, 1 union all
      select 2017, 5, 5 union all
      select 2017, 6, 6
     )
select array_to_string(array_agg(cast(count as string)) over (order by year, month rows between current row and 2 following), ','),
       sum(count) over (order by year, month rows between current row and 2 following)
from t;

将array_agg()用作window函数可能比lead()更麻烦一点。不过再多几个元素，就简单多了。

如何对 google 云 sql 或 python 数据框中的每滚动 3 行进行分组？

How do I group by for every rolling 3 rows in google cloud sql or python dataframe?

python

sql

group-by

dataframe

google-cloud-sql