Accept multiple lambda in groupby list

[TomAugspurger]

We currently don't allow duplicate function names in the list passed too .groupby().agg({'col': [aggfuncs]}). This is painful with multiple lambdas, which all have the name <lambda>

In [1]: import pandas as pd
df
In [2]: df = pd.DataFrame({"A": ['a', 'a'], 'B': [1, 2], 'C': [3, 4]})

In [3]: df.groupby("A").agg({'B': [lambda x: 0, lambda x: 1]})

---------------------------------------------------------------------------
SpecificationError                        Traceback (most recent call last)
~/sandbox/pandas/pandas/core/base.py in _aggregate(self, arg, *args, **kwargs)
    483                 try:
--> 484                     result = _agg(arg, _agg_1dim)
    485                 except SpecificationError:

~/sandbox/pandas/pandas/core/base.py in _agg(arg, func)
    434                 for fname, agg_how in arg.items():
--> 435                     result[fname] = func(fname, agg_how)
    436                 return result

~/sandbox/pandas/pandas/core/base.py in _agg_1dim(name, how, subset)
    417                                              "in aggregation")
--> 418                 return colg.aggregate(how, _level=(_level or 0) + 1)
    419

~/sandbox/pandas/pandas/core/groupby/generic.py in aggregate(self, func_or_funcs, *args, **kwargs)
    771             ret = self._aggregate_multiple_funcs(func_or_funcs,
--> 772                                                  (_level or 0) + 1)
    773         else:

~/sandbox/pandas/pandas/core/groupby/generic.py in _aggregate_multiple_funcs(self, arg, _level)
    834                     'Function names must be unique, found multiple named '
--> 835                     '{}'.format(name))
    836

SpecificationError: Function names must be unique, found multiple named <lambda>

During handling of the above exception, another exception occurred:

SpecificationError                        Traceback (most recent call last)
<ipython-input-3-2aa02bdc2edd> in <module>
----> 1 df.groupby("A").agg({'B': [lambda x: 0, lambda x: 1]})

~/sandbox/pandas/pandas/core/groupby/generic.py in aggregate(self, arg, *args, **kwargs)
   1344     @Appender(_shared_docs['aggregate'])
   1345     def aggregate(self, arg=None, *args, **kwargs):
-> 1346         return super().aggregate(arg, *args, **kwargs)
   1347
   1348     agg = aggregate

~/sandbox/pandas/pandas/core/groupby/generic.py in aggregate(self, func, *args, **kwargs)
    174                             "'(column, aggfunc).")
    175
--> 176         result, how = self._aggregate(func, _level=_level, *args, **kwargs)
    177         if how is None:
    178             return result

~/sandbox/pandas/pandas/core/base.py in _aggregate(self, arg, *args, **kwargs)
    487                     # we are aggregating expecting all 1d-returns
    488                     # but we have 2d
--> 489                     result = _agg(arg, _agg_2dim)
    490
    491             # combine results

~/sandbox/pandas/pandas/core/base.py in _agg(arg, func)
    433                 result = OrderedDict()
    434                 for fname, agg_how in arg.items():
--> 435                     result[fname] = func(fname, agg_how)
    436                 return result
    437

~/sandbox/pandas/pandas/core/base.py in _agg_2dim(name, how)
    424                 colg = self._gotitem(self._selection, ndim=2,
    425                                      subset=obj)
--> 426                 return colg.aggregate(how, _level=None)
    427
    428             def _agg(arg, func):

~/sandbox/pandas/pandas/core/groupby/generic.py in aggregate(self, arg, *args, **kwargs)
   1344     @Appender(_shared_docs['aggregate'])
   1345     def aggregate(self, arg=None, *args, **kwargs):
-> 1346         return super().aggregate(arg, *args, **kwargs)
   1347
   1348     agg = aggregate

~/sandbox/pandas/pandas/core/groupby/generic.py in aggregate(self, func, *args, **kwargs)
    174                             "'(column, aggfunc).")
    175
--> 176         result, how = self._aggregate(func, _level=_level, *args, **kwargs)
    177         if how is None:
    178             return result

~/sandbox/pandas/pandas/core/base.py in _aggregate(self, arg, *args, **kwargs)
    542             return self._aggregate_multiple_funcs(arg,
    543                                                   _level=_level,
--> 544                                                   _axis=_axis), None
    545         else:
    546             result = None

~/sandbox/pandas/pandas/core/base.py in _aggregate_multiple_funcs(self, arg, _level, _axis)
    588                     colg = self._gotitem(col, ndim=1,
    589                                          subset=obj.iloc[:, index])
--> 590                     results.append(colg.aggregate(arg))
    591                     keys.append(col)
    592                 except (TypeError, DataError):

~/sandbox/pandas/pandas/core/groupby/generic.py in aggregate(self, func_or_funcs, *args, **kwargs)
    770             # but not the class list / tuple itself.
    771             ret = self._aggregate_multiple_funcs(func_or_funcs,
--> 772                                                  (_level or 0) + 1)
    773         else:
    774             cyfunc = self._is_cython_func(func_or_funcs)

~/sandbox/pandas/pandas/core/groupby/generic.py in _aggregate_multiple_funcs(self, arg, _level)
    833                 raise SpecificationError(
    834                     'Function names must be unique, found multiple named '
--> 835                     '{}'.format(name))
    836
    837             # reset the cache so that we

SpecificationError: Function names must be unique, found multiple named <lambda>

I propose that we mangle the names somehow

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

In [3]: df.groupby("A").agg({'B': [lambda x: 0, lambda x: 1]})
Out[3]:
         B
  <lambda> <lambda 1>
A
a        0          1

That adds a 1, 2, ... to all subsequent lambdas in the same MI level. It doesn't change the first. Do we want <lambda 0> for the first?

As a side-effect, this enables multiple lambdas per column with the new keyword aggregation

In [4]: df.groupby("A").agg(b=('B', lambda x: 0), c=('B', lambda x: 1))
Out[4]:
   b  c
A
a  0  0

---

I have a WIP started. Will do for 0.25.

[TomAugspurger]

There’s an open PR fixing that.

…

On Aug 21, 2019, at 05:17, Florian Wetschoreck ***@***.***> wrote: Thank you for adding this @TomAugspurger Everything works great for me except for example [4]: df.groupby("A").agg(b=('B', lambda x: 0), c=('B', lambda x: 1)) Which results in the following error: KeyError Traceback (most recent call last) <ipython-input-149-3ff87fe40b34> in <module> 1 df = pd.DataFrame({"A": ['a', 'a'], 'B': [1, 2], 'C': [3, 4]}) 2 #df.groupby("A").agg({'B': [lambda x: 0, lambda x: 1]}) ----> 3 df.groupby("A").agg(b=('B', lambda x: 0), c=('B', lambda x: 1)) 4 #df.groupby("A").agg(b=('B', lambda x: 0), c=('C', lambda x: 1)) /usr/local/lib/python3.7/site-packages/pandas/core/groupby/generic.py in aggregate(self, arg, *args, **kwargs) 1453 @appender(_shared_docs["aggregate"]) 1454 def aggregate(self, arg=None, *args, **kwargs): -> 1455 return super().aggregate(arg, *args, **kwargs) 1456 1457 agg = aggregate /usr/local/lib/python3.7/site-packages/pandas/core/groupby/generic.py in aggregate(self, func, *args, **kwargs) 262 263 if relabeling: --> 264 result = result[order] 265 result.columns = columns 266 /usr/local/lib/python3.7/site-packages/pandas/core/frame.py in __getitem__(self, key) 2979 if is_iterator(key): 2980 key = list(key) -> 2981 indexer = self.loc._convert_to_indexer(key, axis=1, raise_missing=True) 2982 2983 # take() does not accept boolean indexers /usr/local/lib/python3.7/site-packages/pandas/core/indexing.py in _convert_to_indexer(self, obj, axis, is_setter, raise_missing) 1269 # When setting, missing keys are not allowed, even with .loc: 1270 kwargs = {"raise_missing": True if is_setter else raise_missing} -> 1271 return self._get_listlike_indexer(obj, axis, **kwargs)[1] 1272 else: 1273 try: /usr/local/lib/python3.7/site-packages/pandas/core/indexing.py in _get_listlike_indexer(self, key, axis, raise_missing) 1076 1077 self._validate_read_indexer( -> 1078 keyarr, indexer, o._get_axis_number(axis), raise_missing=raise_missing 1079 ) 1080 return keyarr, indexer /usr/local/lib/python3.7/site-packages/pandas/core/indexing.py in _validate_read_indexer(self, key, indexer, axis, raise_missing) 1161 raise KeyError( 1162 "None of [{key}] are in the [{axis}]".format( -> 1163 key=key, axis=self.obj._get_axis_name(axis) 1164 ) 1165 ) KeyError: "None of [MultiIndex([('B', '<lambda>'),\n ('B', '<lambda>')],\n )] are in the [columns]" The new named approach only works for different columns for me. E.g. when I change column B to C in the second aggregation: df.groupby("A").agg(b=('B', lambda x: 0), c=('C', lambda x: 1)) — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.