python : aggregate dataframe values by bin

I have a dataset with that looks like that :

|col A|col B|
   1     20
   3    123
   7     2
     ...

I would like to compute the mean value of col B over each bin of col A.

This would result in a new dataframe containing only one row per bin with :

 | mid value of the col A bin | avg value of col B over that bin |

Solution

As you haven't specified the number of bins and their properties, let me illustrate what you may do with pandas.cut to the example data you provided:

import pandas as pd

# reproduce your example data
df = pd.DataFrame({'col A': [1, 3, 7],
                   'col B': [20, 123, 2]})

# suggest only 2 bins would be proper for 3 rows of data
df['col A bins'] = pd.cut(df['col A'], 
                          bins=2)

Output:

# bins may be labeled as you like, not as automatic interval
    col A   col B   col A bins
0   1       20      (0.994, 4.0]
1   3       123     (0.994, 4.0]
2   7       2       (4.0, 7.0]

Then we may group the initial columns by the new bins, with col A aggregated to median (as from your new column names) and col B to mean, making it look as your expected result by renaming and dropping columns:

df.groupby('col A bins').agg({'col A': 'median',
                              'col B': 'mean'}
                       ).rename(columns={'col A':'mid value of the col A bin',
                                         'col B':'avg value of col B over that bin'}
                       ).reset_index(drop=True)

Output:

    mid value of the col A bin  avg value of col B over that bin
0   2.0                         71.5
1   7.0                         2.0