SQL groupby在熊猫中划分

Question

and now I have such a SQL statement that I would like to know how I can write that in Pandas, maybe using groupBy and apply?:

Give a table with columns of A, B

 Select A, sum(B) / sum(A)
 from table
 group by A;

I am now at

def func(group):
   x = group['B']
   y = group['A']
   return x.sum() / y.sum()

table.groupby('A').apply(func)

This will generate a sequence of numbers without Column A which is used for grouping by on. I would like to have a dataframe as output with A as a separate column also, just like the SQL statement I wrote. Can anyone help me to answer this question?

Thanks!

Answer 1

Is this what you want ?

df=pd.DataFrame({'A':[1,1,3,4],'B':[2,3,4,5]})

def func(group):
   x = group['B']
   y = group['A']
   return x.sum() / y.sum()

df.groupby('A').apply(func).reset_index()


Out[934]: 
   A         0
0  1  2.500000
1  3  1.333333
2  4  1.250000

Answer 2

There's no need for an apply here. It would be a lot faster to groupby , calculate the sum and divide directly, as pandas vectorises these operations.

Borrowing from @Wen's setup, this is how I'd do it -

v = df.groupby('A')[['A', 'B']].sum()
v['B'] /= v['A']
del v['A']

          B
A          
1  2.500000
3  1.333333
4  1.250000