问题
I have a bunch of data stored in vals. The indices are monotonic, but not continuous. I'm attempting to do some analysis on histograms of the data, so I've created the following structure:
hist = pd.DataFrame(vals)
hist['bins'] = pd.cut(vals, 100)
This is data taken from an experimental instrument and I know that some of the bins have only 1 or 2 counts in them, which I'm trying to remove. I've tried using groupby as follows and get the following error (Full traceback included at the end of the note):
hist.groupby('bins').describe()
AttributeError: 'Categorical' object has no attribute 'flags'
However, when I do the following, the error does not show up and I get the expected result:
In[]: hist.index = hist.bins
In[]: hist['bins'] = hist.index
In[]: desc = hist.groupby('bins').describe()
In[]: desc.index.names = ['bins', 'describe']
Out[]: **describe with MultiIndex for rows.**
If I don't include the second line hist['bins'] = hist.index, I still get an AttributeError: 'Categorical' object has no attribute 'flags' and to the best that I can tell, the traceback is identical.
Can someone explain what the flags are and why they only seem to work when I set the index to bins and then replace the bins by the version stored in the index?
My end goal is to remove the data for bins with counts <= 6. If someone has an easier workaround than the way I'm going after it, I'd also be grateful.
---------------------------------------------------------------------------
AttributeError Traceback (most recent call last)
<ipython-input-11-f606a051f2e4> in <module>()
----> 1 hist.groupby('bins').describe()
C:\Users\balterma\AppData\Local\Enthought\Canopy\App\appdata\canopy-1.4.1.1975.win-x86_64\lib\site-packages\IPython\core\displayhook.pyc in __call__(self, result)
245 self.start_displayhook()
246 self.write_output_prompt()
--> 247 format_dict, md_dict = self.compute_format_data(result)
248 self.write_format_data(format_dict, md_dict)
249 self.update_user_ns(result)
C:\Users\balterma\AppData\Local\Enthought\Canopy\App\appdata\canopy-1.4.1.1975.win-x86_64\lib\site-packages\IPython\core\displayhook.pyc in compute_format_data(self, result)
155
156 """
--> 157 return self.shell.display_formatter.format(result)
158
159 def write_format_data(self, format_dict, md_dict=None):
C:\Users\balterma\AppData\Local\Enthought\Canopy\App\appdata\canopy-1.4.1.1975.win-x86_64\lib\site-packages\IPython\core\formatters.pyc in format(self, obj, include, exclude)
150 md = None
151 try:
--> 152 data = formatter(obj)
153 except:
154 # FIXME: log the exception
C:\Users\balterma\AppData\Local\Enthought\Canopy\App\appdata\canopy-1.4.1.1975.win-x86_64\lib\site-packages\IPython\core\formatters.pyc in __call__(self, obj)
479 type_pprinters=self.type_printers,
480 deferred_pprinters=self.deferred_printers)
--> 481 printer.pretty(obj)
482 printer.flush()
483 return stream.getvalue()
C:\Users\balterma\AppData\Local\Enthought\Canopy\App\appdata\canopy-1.4.1.1975.win-x86_64\lib\site-packages\IPython\lib\pretty.pyc in pretty(self, obj)
360 if callable(meth):
361 return meth(obj, self, cycle)
--> 362 return _default_pprint(obj, self, cycle)
363 finally:
364 self.end_group()
C:\Users\balterma\AppData\Local\Enthought\Canopy\App\appdata\canopy-1.4.1.1975.win-x86_64\lib\site-packages\IPython\lib\pretty.pyc in _default_pprint(obj, p, cycle)
480 if getattr(klass, '__repr__', None) not in _baseclass_reprs:
481 # A user-provided repr.
--> 482 p.text(repr(obj))
483 return
484 p.begin_group(1, '<')
C:\Users\balterma\AppData\Local\Enthought\Canopy\User\lib\site-packages\pandas\core\base.pyc in __repr__(self)
62 Yields Bytestring in Py2, Unicode String in py3.
63 """
---> 64 return str(self)
65
66
C:\Users\balterma\AppData\Local\Enthought\Canopy\User\lib\site-packages\pandas\core\base.pyc in __str__(self)
42 if compat.PY3:
43 return self.__unicode__()
---> 44 return self.__bytes__()
45
46 def __bytes__(self):
C:\Users\balterma\AppData\Local\Enthought\Canopy\User\lib\site-packages\pandas\core\base.pyc in __bytes__(self)
54
55 encoding = get_option("display.encoding")
---> 56 return self.__unicode__().encode(encoding, 'replace')
57
58 def __repr__(self):
C:\Users\balterma\AppData\Local\Enthought\Canopy\User\lib\site-packages\pandas\core\frame.pyc in __unicode__(self)
507 width = None
508 self.to_string(buf=buf, max_rows=max_rows, max_cols=max_cols,
--> 509 line_width=width, show_dimensions=show_dimensions)
510
511 return buf.getvalue()
C:\Users\balterma\AppData\Local\Enthought\Canopy\User\lib\site-packages\pandas\core\frame.pyc in to_string(self, buf, columns, col_space, colSpace, header, index, na_rep, formatters, float_format, sparsify, index_names, justify, line_width, max_rows, max_cols, show_dimensions)
1340 max_rows=max_rows,
1341 max_cols=max_cols,
-> 1342 show_dimensions=show_dimensions)
1343 formatter.to_string()
1344
C:\Users\balterma\AppData\Local\Enthought\Canopy\User\lib\site-packages\pandas\core\format.pyc in __init__(self, frame, buf, columns, col_space, header, index, na_rep, formatters, justify, float_format, sparsify, index_names, line_width, max_rows, max_cols, show_dimensions, **kwds)
345 self.columns = frame.columns
346
--> 347 self._chk_truncate()
348
349 def _chk_truncate(self):
C:\Users\balterma\AppData\Local\Enthought\Canopy\User\lib\site-packages\pandas\core\format.pyc in _chk_truncate(self)
410 else:
411 row_num = max_rows_adj // 2
--> 412 frame = concat((frame.iloc[:row_num, :], frame.iloc[-row_num:, :]))
413 self.tr_row_num = row_num
414
C:\Users\balterma\AppData\Local\Enthought\Canopy\User\lib\site-packages\pandas\tools\merge.pyc in concat(objs, axis, join, join_axes, ignore_index, keys, levels, names, verify_integrity, copy)
752 keys=keys, levels=levels, names=names,
753 verify_integrity=verify_integrity,
--> 754 copy=copy)
755 return op.get_result()
756
C:\Users\balterma\AppData\Local\Enthought\Canopy\User\lib\site-packages\pandas\tools\merge.pyc in __init__(self, objs, axis, join, join_axes, keys, levels, names, ignore_index, verify_integrity, copy)
884 self.copy = copy
885
--> 886 self.new_axes = self._get_new_axes()
887
888 def get_result(self):
C:\Users\balterma\AppData\Local\Enthought\Canopy\User\lib\site-packages\pandas\tools\merge.pyc in _get_new_axes(self)
957 new_axes[i] = ax
958
--> 959 new_axes[self.axis] = self._get_concat_axis()
960 return new_axes
961
C:\Users\balterma\AppData\Local\Enthought\Canopy\User\lib\site-packages\pandas\tools\merge.pyc in _get_concat_axis(self)
1009
1010 if self.keys is None:
-> 1011 concat_axis = _concat_indexes(indexes)
1012 else:
1013 concat_axis = _make_concat_multiindex(indexes, self.keys,
C:\Users\balterma\AppData\Local\Enthought\Canopy\User\lib\site-packages\pandas\tools\merge.pyc in _concat_indexes(indexes)
1027
1028 def _concat_indexes(indexes):
-> 1029 return indexes[0].append(indexes[1:])
1030
1031
C:\Users\balterma\AppData\Local\Enthought\Canopy\User\lib\site-packages\pandas\core\index.pyc in append(self, other)
4603 arrays = []
4604 for i in range(self.nlevels):
-> 4605 label = self.get_level_values(i)
4606 appended = [o.get_level_values(i) for o in other]
4607 arrays.append(label.append(appended))
C:\Users\balterma\AppData\Local\Enthought\Canopy\User\lib\site-packages\pandas\core\index.pyc in get_level_values(self, level)
4239 unique = self.levels[num] # .values
4240 labels = self.labels[num]
-> 4241 filled = com.take_1d(unique.values, labels, fill_value=unique._na_value)
4242 values = unique._simple_new(filled, self.names[num],
4243 freq=getattr(unique, 'freq', None),
C:\Users\balterma\AppData\Local\Enthought\Canopy\User\lib\site-packages\pandas\core\common.pyc in take_nd(arr, indexer, axis, out, fill_value, mask_info, allow_fill)
829 out_shape[axis] = len(indexer)
830 out_shape = tuple(out_shape)
--> 831 if arr.flags.f_contiguous and axis == arr.ndim - 1:
832 # minor tweak that can make an order-of-magnitude difference
833 # for dataframes initialized directly from 2-d ndarrays
AttributeError: 'Categorical' object has no attribute 'flags'
回答1:
This looks to be be a bug with Categorical data that will be corrected in version 0.17.0 (issue here).
In the meantime, you could just cast the category to an object dtype - this is what was happening when you assigned to the index and back.
df['bins'] = df['bins'].astype(str)
来源：https://stackoverflow.com/questions/31999862/pandas-groupby-and-describe-flags-attributeerror