用于更复杂的sql查询的pandas语法

Question

我正在寻找pandas语法来对Pandas DataFrame进行以下聚合。我找不到如何在pandas中完成以下sql查询的示例。

#sum and divide
select
 click,
 ctr,
 SUM(click)/SUM(imp) as ctr
from mytable
group by website

#normalize each subgroup 
select
 imp,
 imp/SUM(imp) Over (partition website) as n_imp
from mytable

Answer 1

SQL：

#normalize each subgroup 
select
 imp,
 imp/SUM(imp) Over (partition website) as n_imp
from mytable

熊猫：

df[['website','imp']].assign(n_imp=df['imp']/df.groupby('website')['imp'].transform('sum'))