"Duplicated level name" when using groupby with two different attributes of a datetime

[mattharrison]

Code Sample, a copy-pastable example if possible

from io import StringIO
import pandas as pd

data = '''Date,Amount
10/30/2010,54
11/20/2010,53'''
df = pd.read_csv(StringIO(data), parse_dates=[0])

df.groupby([df.Date.dt.year, df.Date.dt.month]).mean()

Problem description

This worked prior to 0.23 and seems like a regression. I now get the following error:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
~/.env/36/lib/python3.6/site-packages/pandas/core/groupby/groupby.py in _cython_agg_blocks(self, how, alt, numeric_only, min_count)
   4017                 result, _ = self.grouper.aggregate(
-> 4018                     block.values, how, axis=agg_axis, min_count=min_count)
   4019             except NotImplementedError:

~/.env/36/lib/python3.6/site-packages/pandas/core/groupby/groupby.py in aggregate(self, values, how, axis, min_count)
   2626         return self._cython_operation('aggregate', values, how, axis,
-> 2627                                       min_count=min_count)
   2628 

~/.env/36/lib/python3.6/site-packages/pandas/core/groupby/groupby.py in _cython_operation(self, kind, values, how, axis, min_count, **kwargs)
   2532                                           "supported for the 'how' argument")
-> 2533             out_shape = (self.ngroups,) + values.shape[1:]
   2534 

pandas/_libs/properties.pyx in pandas._libs.properties.CachedProperty.__get__()

~/.env/36/lib/python3.6/site-packages/pandas/core/groupby/groupby.py in ngroups(self)
   2361     def ngroups(self):
-> 2362         return len(self.result_index)
   2363 

pandas/_libs/properties.pyx in pandas._libs.properties.CachedProperty.__get__()

~/.env/36/lib/python3.6/site-packages/pandas/core/groupby/groupby.py in result_index(self)
   2380                             verify_integrity=False,
-> 2381                             names=self.names)
   2382         return result

~/.env/36/lib/python3.6/site-packages/pandas/core/indexes/multi.py in __new__(cls, levels, labels, sortorder, names, dtype, copy, name, verify_integrity, _set_identity)
    231             # handles name validation
--> 232             result._set_names(names)
    233 

~/.env/36/lib/python3.6/site-packages/pandas/core/indexes/multi.py in _set_names(self, names, level, validate)
    694                         'level {}, is already used for level '
--> 695                         '{}.'.format(name, l, used[name]))
    696 

ValueError: Duplicated level name: "Date", assigned to level 1, is already used for level 0.

During handling of the above exception, another exception occurred:

UnboundLocalError                         Traceback (most recent call last)
~/.env/36/lib/python3.6/site-packages/pandas/core/groupby/groupby.py in mean(self, *args, **kwargs)
   1305         try:
-> 1306             return self._cython_agg_general('mean', **kwargs)
   1307         except GroupByError:

~/.env/36/lib/python3.6/site-packages/pandas/core/groupby/groupby.py in _cython_agg_general(self, how, alt, numeric_only, min_count)
   3973         new_items, new_blocks = self._cython_agg_blocks(
-> 3974             how, alt=alt, numeric_only=numeric_only, min_count=min_count)
   3975         return self._wrap_agged_blocks(new_items, new_blocks)

~/.env/36/lib/python3.6/site-packages/pandas/core/groupby/groupby.py in _cython_agg_blocks(self, how, alt, numeric_only, min_count)
   4038                 # see if we can cast the block back to the original dtype
-> 4039                 result = block._try_coerce_and_cast_result(result)
   4040                 newb = block.make_block(result)

UnboundLocalError: local variable 'result' referenced before assignment

During handling of the above exception, another exception occurred:

ValueError                                Traceback (most recent call last)
<ipython-input-13-4b644a75a9ce> in <module>()
     10 df.dtypes
     11 
---> 12 df.groupby([df.Date.dt.year, df.Date.dt.month]).mean()

~/.env/36/lib/python3.6/site-packages/pandas/core/groupby/groupby.py in mean(self, *args, **kwargs)
   1310             with _group_selection_context(self):
   1311                 f = lambda x: x.mean(axis=self.axis, **kwargs)
-> 1312                 return self._python_agg_general(f)
   1313 
   1314     @Substitution(name='groupby')

~/.env/36/lib/python3.6/site-packages/pandas/core/groupby/groupby.py in _python_agg_general(self, func, *args, **kwargs)
   1086                 output[name] = self._try_cast(values[mask], result)
   1087 
-> 1088         return self._wrap_aggregated_output(output)
   1089 
   1090     def _wrap_applied_output(self, *args, **kwargs):

~/.env/36/lib/python3.6/site-packages/pandas/core/groupby/groupby.py in _wrap_aggregated_output(self, output, names)
   4728             result = result._consolidate()
   4729         else:
-> 4730             index = self.grouper.result_index
   4731             result = DataFrame(output, index=index, columns=output_keys)
   4732 

pandas/_libs/properties.pyx in pandas._libs.properties.CachedProperty.__get__()

~/.env/36/lib/python3.6/site-packages/pandas/core/groupby/groupby.py in result_index(self)
   2379                             labels=labels,
   2380                             verify_integrity=False,
-> 2381                             names=self.names)
   2382         return result
   2383 

~/.env/36/lib/python3.6/site-packages/pandas/core/indexes/multi.py in __new__(cls, levels, labels, sortorder, names, dtype, copy, name, verify_integrity, _set_identity)
    230         if names is not None:
    231             # handles name validation
--> 232             result._set_names(names)
    233 
    234         if sortorder is not None:

~/.env/36/lib/python3.6/site-packages/pandas/core/indexes/multi.py in _set_names(self, names, level, validate)
    693                         'Duplicated level name: "{}", assigned to '
    694                         'level {}, is already used for level '
--> 695                         '{}.'.format(name, l, used[name]))
    696 
    697             self.levels[l].rename(name, inplace=True)

ValueError: Duplicated level name: "Date", assigned to level 1, is already used for level 0.

Expected Output

           Amount
Date Date        
2010 10        54
     11        53

Output of pd.show_versions()

INSTALLED VERSIONS ------------------ commit: None python: 3.6.4.final.0 python-bits: 64 OS: Darwin OS-release: 16.5.0 machine: x86_64 processor: i386 byteorder: little LC_ALL: None LANG: en_US.UTF-8 LOCALE: en_US.UTF-8

pandas: 0.23.0
pytest: 3.0.6
pip: 10.0.1
setuptools: 39.2.0
Cython: 0.27.3
numpy: 1.14.3
scipy: 1.1.0
pyarrow: None
xarray: None
IPython: 6.2.1
sphinx: 1.7.5
patsy: 0.5.0
dateutil: 2.7.3
pytz: 2018.4
blosc: None
bottleneck: None
tables: None
numexpr: None
feather: None
matplotlib: 2.2.2
openpyxl: 2.4.7
xlrd: 1.0.0
xlwt: 1.2.0
xlsxwriter: None
lxml: 3.7.2
bs4: 4.6.0
html5lib: 0.9999999
sqlalchemy: 1.2.5
pymysql: None
psycopg2: 2.7.4 (dt dec pq3 ext lo64)
jinja2: 2.10
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None

[jschendel]

Thanks for the report. This is a dupe of #21075 (see comments for a workaround), which was closed in favor of #19029.

[mattharrison]

(My bad with the search fu). It seems to me that pulling off a date time attribute should change the column name. I see the pr that tacks on 1,2,etc, but wouldn't it be better to tack on "_month", "_hour", etc?

[jschendel]

No worries. Would rather have two dupes than no issue being reported at all!

For the workaround I was referring #21075 (comment), or in you case:

df.groupby([df.Date.dt.year.rename('year'), df.Date.dt.month.rename(month')]).mean()

I suspect the PR is tacking on 0,1,... since that handles things most generically, and it corresponds to the level enumeration of a MultiIndex.

As to why something like df.Date.dt.year doesn't automatically change the name? I don't know for sure, but I suspect it's a combination of:

1. I don't know of a well agreed upon convention of how to rename (rename entirely, add suffixes, etc.)
2. There's a rename method that's easily chainable to explicitly accommodate any desired renaming
3. Probably best to retain existing behavior in the absence of a clearly better alternative

That being said, feel free to open a new issue if you feel strongly about automatically renaming these things.