pivot_table throws exception with multiple aggregations per column and margins=True

When applying multiple aggregations to columns and setting margins=True I receive a KeyError. I believe that because multiple aggregations are applied the columns become a MultiIndex, which is unexpected when computing margins.

>>> import pandas as pd
>>> import numpy as np
>>> import random


>>> df = pd.DataFrame({'random1': [random.random() for i in range(10)],
                       'random2': [random.random() for i in range(10)],
                       'type': ['duck', 'bird']*5},
                index=range(10,20))

>>> df.pivot_table(index='type', 
                   aggfunc={'random1': [np.median, np.mean], 
                            'random2': np.sum}, 
                   margins=True)

KeyError                                  Traceback (most recent call last)
<ipython-input> in <module>()
      3                aggfunc={'random1': [np.median, np.mean], 
      4                         'random2': np.sum}, 
----> 5                margins=True)

/pandas/util/decorators.pyc in wrapper(*args, **kwargs)
     86                 else:
     87                     kwargs[new_arg_name] = new_arg_value
---> 88             return func(*args, **kwargs)
     89         return wrapper
     90     return _deprecate_kwarg

/pandas/util/decorators.pyc in wrapper(*args, **kwargs)
     86                 else:
     87                     kwargs[new_arg_name] = new_arg_value
---> 88             return func(*args, **kwargs)
     89         return wrapper
     90     return _deprecate_kwarg

/pandas/tools/pivot.pyc in pivot_table(data, values, index, columns, aggfunc, fill_value, margins, dropna)
    145     if margins:
    146         table = _add_margins(table, data, values, rows=index,
--> 147                              cols=columns, aggfunc=aggfunc)
    148 
    149     # discard the top level

/pandas/tools/pivot.pyc in _add_margins(table, data, values, rows, cols, aggfunc)
    189             row_margin[k] = grand_margin[k]
    190         else:
--> 191             row_margin[k] = grand_margin[k[0]]
    192 
    193     margin_dummy = DataFrame(row_margin, columns=[key]).T

KeyError: 'random1'

The exact same error occurs with

aggfunc={'random1': {'median': np.median, 'mean': np.mean}, 
         'random2': np.sum}

These errors do not occur when margins=False This is using pandas version 0.15.2. I have not seen any changes which would lead me to believe it is be fixed in master.

Happy to create a PR, although I haven’t done one before.

[stefansimik]

Hi @tmo, If you could check into this, it would be really great. I would like to do it, but is is above my experience level to fix this in pandas codebase.

[lukaszkiszka]

@tmo go ahead - I was going to look on this in future - in a few weeks. If you will found resolution I will be glad when you will add me as reviewer.