Python sum dataframe column in batches

I have a 100 rows DataFrame and I would like to sum together every X rows from some column.

I tried to using rolling and cumsum but it does not helped me.

lets say this is my column:

np.arange(1,100)

What I am trying to do is to sum the numbers from 1 to 10, 11 to 20 and so on (just example when X=10).

Any other solutions?

Thanks!

Solution

assuming that your index is an auto-increment starting at 0, you could do the following:

import numpy as np
import pandas as pd
df = pd.DataFrame({'a': np.arange(1,100)})
batch_size = 10
df.groupby((df.index // batch_size).rename('group_number')).a.sum().reset_index()

This outputs the following dataframe:

   group_number    a
0             0   55
1             1  155
2             2  255
3             3  355
4             4  455
5             5  555
6             6  655
7             7  755
8             8  855
9             9  855

the grouping is done by the expression df.index // batch_size, which will split all, except perhaps the last group of rows into groups of consecutive rows of size batch_size. The last group may have fewer than batch_size rows.

we know that

the sum of all integers between 1 & 10 is 55, (1 + 10) * (10 / 2)
the sum of all integers between 11 & 20 is 155, (11 + 20) * (10 / 2)
the other group-sums also check out with the Gauss's Summation Technique.

so my solution seems to checkout.