Geek Answers Handbook

Aggregate Group with Layered Columns

I have a grouped DataFrame that I want to combine with a dictionary of functions that should appear in specific columns. It is easy for single-level columns: groups.agg({'colname': <function>}). However, I am struggling for this to work with multi-level columns, of which I only want to reference one level.

Here is an example.

Allows you to make some sample data:

import itertools
import pandas as pd

lev1 = ['foo', 'bar', 'baz']
lev2 = list('abc')

n = 6

df = pd.DataFrame({k: np.random.randn(n) for k in itertools.product(lev1,lev2)}, 
                  index=pd.DatetimeIndex(start='2015-01-01', periods=n, freq='11D'))

It looks like this:

             bar               baz               foo            
               a     b     c     a     b     c     a     b     c
2015-01-01 -1.11  2.12 -1.00  0.18  0.14  1.24  0.73  0.06  3.66
2015-01-12 -1.43  0.75  0.38  0.04 -0.33 -0.42  1.00 -1.63 -1.35
2015-01-23  0.01 -1.70 -1.39  0.59 -1.10 -1.17 -1.51 -0.54 -1.11
2015-02-03  0.93  0.70 -0.12  1.07 -0.97 -0.45 -0.19  0.11 -0.79
2015-02-14  0.30  0.49  0.60 -0.28 -0.38  1.11  0.15  0.78 -0.58
2015-02-25 -0.26  0.51  0.82  0.05 -1.45  0.14  0.53 -0.33 -1.35

And grouping by month:

groups = df.groupby(pd.TimeGrouper('MS'))

Define some functions based on the top level in columns:

funcs = {'bar': np.sum, 'baz': np.mean, 'foo': np.min}

However, execution groups.agg(funcs)results in a KeyError as it expects a key for each level, which makes sense.

This works for example:

groups.agg({('bar', 'a'): np.mean})

                 bar
                   a
2015-01-01 -0.845554
2015-02-01  0.324897

But I do not want to indicate each key in the second level. Therefore, I am looking for something that will work as follows:

groups.agg({('bar', slice(None)): np.mean})

, slice .

:

def multifunc(group):

    func = funcs[group.name[0]]        
    return func(group)

groups.agg(multifunc)

"". , agg. / , .

+4
1

, . , dict:

result = groups.agg(
    {(k1, k2): funcs[k1] for k1, k2 in itertools.product(lev1,lev2)})

import itertools
import numpy as np
import pandas as pd

lev1 = ['foo', 'bar', 'baz']
lev2 = list('abc')

n = 6

df = pd.DataFrame(
    {k: np.random.randn(n) for k in itertools.product(lev1,lev2)}, 
    index=pd.DatetimeIndex(start='2015-01-01', periods=n, freq='11D'))
groups = df.groupby(pd.TimeGrouper('MS'))
funcs = {'bar': np.sum, 'baz': np.mean, 'foo': np.min}
result = groups.agg(
    {(k1, k2): funcs[k1] for k1, k2 in itertools.product(lev1,lev2)})
result = result.sortlevel(axis=1)
print(result)


                 bar                           baz                      \
                   a         b         c         a         b         c   
2015-01-01 -2.144890  1.075044  1.038169 -0.460649 -0.309966 -0.211147   
2015-02-01  1.313744  0.247171  1.049129 -0.174827 -0.437982 -0.196427   

                 foo                      
                   a         b         c  
2015-01-01 -1.358973 -1.846916 -0.896234  
2015-02-01 -1.354953 -0.699607  0.288214
+2

More articles:


All Articles