Duplicate Labels#
Index objects are not required to be unique; you can have duplicate row
or column labels. This may be a bit confusing at first. If you’re familiar with
SQL, you know that row labels are similar to a primary key on a table, and you
would never want duplicates in a SQL table. But one of pandas’ roles is to clean
messy, real-world data before it goes to some downstream system. And real-world
data has duplicates, even in fields that are supposed to be unique.
This section describes how duplicate labels change the behavior of certain operations, and how prevent duplicates from arising during operations, or to detect them if they do.
In [1]: import pandas as pd
In [2]: import numpy as np
Consequences of Duplicate Labels#
Some pandas methods (Series.reindex() for example) just don’t work with
duplicates present. The output can’t be determined, and so pandas raises.
In [3]: s1 = pd.Series([0, 1, 2], index=["a", "b", "b"])
In [4]: s1.reindex(["a", "b", "c"])
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
Cell In[4], line 1
----> 1 s1.reindex(["a", "b", "c"])
File ~/work/pandas/pandas/pandas/core/series.py:5164, in Series.reindex(self, index, axis, method, copy, level, fill_value, limit, tolerance)
5147 @doc(
5148 NDFrame.reindex, # type: ignore[has-type]
5149 klass=_shared_doc_kwargs["klass"],
(...)
5162 tolerance=None,
5163 ) -> Series:
-> 5164 return super().reindex(
5165 index=index,
5166 method=method,
5167 copy=copy,
5168 level=level,
5169 fill_value=fill_value,
5170 limit=limit,
5171 tolerance=tolerance,
5172 )
File ~/work/pandas/pandas/pandas/core/generic.py:5629, in NDFrame.reindex(self, labels, index, columns, axis, method, copy, level, fill_value, limit, tolerance)
5626 return self._reindex_multi(axes, copy, fill_value)
5628 # perform the reindex on the axes
-> 5629 return self._reindex_axes(
5630 axes, level, limit, tolerance, method, fill_value, copy
5631 ).__finalize__(self, method="reindex")
File ~/work/pandas/pandas/pandas/core/generic.py:5652, in NDFrame._reindex_axes(self, axes, level, limit, tolerance, method, fill_value, copy)
5649 continue
5651 ax = self._get_axis(a)
-> 5652 new_index, indexer = ax.reindex(
5653 labels, level=level, limit=limit, tolerance=tolerance, method=method
5654 )
5656 axis = self._get_axis_number(a)
5657 obj = obj._reindex_with_indexers(
5658 {axis: [new_index, indexer]},
5659 fill_value=fill_value,
5660 copy=copy,
5661 allow_dups=False,
5662 )
File ~/work/pandas/pandas/pandas/core/indexes/base.py:4436, in Index.reindex(self, target, method, level, limit, tolerance)
4433 raise ValueError("cannot handle a non-unique multi-index!")
4434 elif not self.is_unique:
4435 # GH#42568
-> 4436 raise ValueError("cannot reindex on an axis with duplicate labels")
4437 else:
4438 indexer, _ = self.get_indexer_non_unique(target)
ValueError: cannot reindex on an axis with duplicate labels
Other methods, like indexing, can give very surprising results. Typically
indexing with a scalar will reduce dimensionality. Slicing a DataFrame
with a scalar will return a Series. Slicing a Series with a scalar will
return a scalar. But with duplicates, this isn’t the case.
In [5]: df1 = pd.DataFrame([[0, 1, 2], [3, 4, 5]], columns=["A", "A", "B"])
In [6]: df1
Out[6]:
A A B
0 0 1 2
1 3 4 5
We have duplicates in the columns. If we slice 'B', we get back a Series
In [7]: df1["B"] # a series
Out[7]:
0 2
1 5
Name: B, dtype: int64
But slicing 'A' returns a DataFrame
In [8]: df1["A"] # a DataFrame
Out[8]:
A A
0 0 1
1 3 4
This applies to row labels as well
In [9]: df2 = pd.DataFrame({"A": [0, 1, 2]}, index=["a", "a", "b"])
In [10]: df2
Out[10]:
A
a 0
a 1
b 2
In [11]: df2.loc["b", "A"] # a scalar
Out[11]: 2
In [12]: df2.loc["a", "A"] # a Series
Out[12]:
a 0
a 1
Name: A, dtype: int64
Duplicate Label Detection#
You can check whether an Index (storing the row or column labels) is
unique with Index.is_unique:
In [13]: df2
Out[13]:
A
a 0
a 1
b 2
In [14]: df2.index.is_unique
Out[14]: False
In [15]: df2.columns.is_unique
Out[15]: True
Note
Checking whether an index is unique is somewhat expensive for large datasets. pandas does cache this result, so re-checking on the same index is very fast.
Index.duplicated() will return a boolean ndarray indicating whether a
label is repeated.
In [16]: df2.index.duplicated()
Out[16]: array([False, True, False])
Which can be used as a boolean filter to drop duplicate rows.
In [17]: df2.loc[~df2.index.duplicated(), :]
Out[17]:
A
a 0
b 2
If you need additional logic to handle duplicate labels, rather than just
dropping the repeats, using groupby() on the index is a common
trick. For example, we’ll resolve duplicates by taking the average of all rows
with the same label.
In [18]: df2.groupby(level=0).mean()
Out[18]:
A
a 0.5
b 2.0
Disallowing Duplicate Labels#
Added in version 1.2.0.
As noted above, handling duplicates is an important feature when reading in raw
data. That said, you may want to avoid introducing duplicates as part of a data
processing pipeline (from methods like pandas.concat(),
rename(), etc.). Both Series and DataFrame
disallow duplicate labels by calling .set_flags(allows_duplicate_labels=False).
(the default is to allow them). If there are duplicate labels, an exception
will be raised.
In [19]: pd.Series([0, 1, 2], index=["a", "b", "b"]).set_flags(allows_duplicate_labels=False)
---------------------------------------------------------------------------
DuplicateLabelError Traceback (most recent call last)
Cell In[19], line 1
----> 1 pd.Series([0, 1, 2], index=["a", "b", "b"]).set_flags(allows_duplicate_labels=False)
File ~/work/pandas/pandas/pandas/core/generic.py:508, in NDFrame.set_flags(self, copy, allows_duplicate_labels)
506 df = self.copy(deep=copy and not using_copy_on_write())
507 if allows_duplicate_labels is not None:
--> 508 df.flags["allows_duplicate_labels"] = allows_duplicate_labels
509 return df
File ~/work/pandas/pandas/pandas/core/flags.py:109, in Flags.__setitem__(self, key, value)
107 if key not in self._keys:
108 raise ValueError(f"Unknown flag {key}. Must be one of {self._keys}")
--> 109 setattr(self, key, value)
File ~/work/pandas/pandas/pandas/core/flags.py:96, in Flags.allows_duplicate_labels(self, value)
94 if not value:
95 for ax in obj.axes:
---> 96 ax._maybe_check_unique()
98 self._allows_duplicate_labels = value
File ~/work/pandas/pandas/pandas/core/indexes/base.py:716, in Index._maybe_check_unique(self)
713 duplicates = self._format_duplicate_message()
714 msg += f"\n{duplicates}"
--> 716 raise DuplicateLabelError(msg)
DuplicateLabelError: Index has duplicates.
positions
label
b [1, 2]
This applies to both row and column labels for a DataFrame
In [20]: pd.DataFrame([[0, 1, 2], [3, 4, 5]], columns=["A", "B", "C"],).set_flags(
....: allows_duplicate_labels=False
....: )
....:
Out[20]:
A B C
0 0 1 2
1 3 4 5
This attribute can be checked or set with allows_duplicate_labels,
which indicates whether that object can have duplicate labels.
In [21]: df = pd.DataFrame({"A": [0, 1, 2, 3]}, index=["x", "y", "X", "Y"]).set_flags(
....: allows_duplicate_labels=False
....: )
....:
In [22]: df
Out[22]:
A
x 0
y 1
X 2
Y 3
In [23]: df.flags.allows_duplicate_labels
Out[23]: False
DataFrame.set_flags() can be used to return a new DataFrame with attributes
like allows_duplicate_labels set to some value
In [24]: df2 = df.set_flags(allows_duplicate_labels=True)
In [25]: df2.flags.allows_duplicate_labels
Out[25]: True
The new DataFrame returned is a view on the same data as the old DataFrame.
Or the property can just be set directly on the same object
In [26]: df2.flags.allows_duplicate_labels = False
In [27]: df2.flags.allows_duplicate_labels
Out[27]: False
When processing raw, messy data you might initially read in the messy data (which potentially has duplicate labels), deduplicate, and then disallow duplicates going forward, to ensure that your data pipeline doesn’t introduce duplicates.
>>> raw = pd.read_csv("...")
>>> deduplicated = raw.groupby(level=0).first() # remove duplicates
>>> deduplicated.flags.allows_duplicate_labels = False # disallow going forward
Setting allows_duplicate_labels=False on a Series or DataFrame with duplicate
labels or performing an operation that introduces duplicate labels on a Series or
DataFrame that disallows duplicates will raise an
errors.DuplicateLabelError.
In [28]: df.rename(str.upper)
---------------------------------------------------------------------------
DuplicateLabelError Traceback (most recent call last)
Cell In[28], line 1
----> 1 df.rename(str.upper)
File ~/work/pandas/pandas/pandas/core/frame.py:5774, in DataFrame.rename(self, mapper, index, columns, axis, copy, inplace, level, errors)
5643 def rename(
5644 self,
5645 mapper: Renamer | None = None,
(...)
5653 errors: IgnoreRaise = "ignore",
5654 ) -> DataFrame | None:
5655 """
5656 Rename columns or index labels.
5657
(...)
5772 4 3 6
5773 """
-> 5774 return super()._rename(
5775 mapper=mapper,
5776 index=index,
5777 columns=columns,
5778 axis=axis,
5779 copy=copy,
5780 inplace=inplace,
5781 level=level,
5782 errors=errors,
5783 )
File ~/work/pandas/pandas/pandas/core/generic.py:1140, in NDFrame._rename(self, mapper, index, columns, axis, copy, inplace, level, errors)
1138 return None
1139 else:
-> 1140 return result.__finalize__(self, method="rename")
File ~/work/pandas/pandas/pandas/core/generic.py:6281, in NDFrame.__finalize__(self, other, method, **kwargs)
6274 if other.attrs:
6275 # We want attrs propagation to have minimal performance
6276 # impact if attrs are not used; i.e. attrs is an empty dict.
6277 # One could make the deepcopy unconditionally, but a deepcopy
6278 # of an empty dict is 50x more expensive than the empty check.
6279 self.attrs = deepcopy(other.attrs)
-> 6281 self.flags.allows_duplicate_labels = other.flags.allows_duplicate_labels
6282 # For subclasses using _metadata.
6283 for name in set(self._metadata) & set(other._metadata):
File ~/work/pandas/pandas/pandas/core/flags.py:96, in Flags.allows_duplicate_labels(self, value)
94 if not value:
95 for ax in obj.axes:
---> 96 ax._maybe_check_unique()
98 self._allows_duplicate_labels = value
File ~/work/pandas/pandas/pandas/core/indexes/base.py:716, in Index._maybe_check_unique(self)
713 duplicates = self._format_duplicate_message()
714 msg += f"\n{duplicates}"
--> 716 raise DuplicateLabelError(msg)
DuplicateLabelError: Index has duplicates.
positions
label
X [0, 2]
Y [1, 3]
This error message contains the labels that are duplicated, and the numeric positions
of all the duplicates (including the “original”) in the Series or DataFrame
Duplicate Label Propagation#
In general, disallowing duplicates is “sticky”. It’s preserved through operations.
In [29]: s1 = pd.Series(0, index=["a", "b"]).set_flags(allows_duplicate_labels=False)
In [30]: s1
Out[30]:
a 0
b 0
dtype: int64
In [31]: s1.head().rename({"a": "b"})
---------------------------------------------------------------------------
DuplicateLabelError Traceback (most recent call last)
Cell In[31], line 1
----> 1 s1.head().rename({"a": "b"})
File ~/work/pandas/pandas/pandas/core/series.py:5101, in Series.rename(self, index, axis, copy, inplace, level, errors)
5094 axis = self._get_axis_number(axis)
5096 if callable(index) or is_dict_like(index):
5097 # error: Argument 1 to "_rename" of "NDFrame" has incompatible
5098 # type "Union[Union[Mapping[Any, Hashable], Callable[[Any],
5099 # Hashable]], Hashable, None]"; expected "Union[Mapping[Any,
5100 # Hashable], Callable[[Any], Hashable], None]"
-> 5101 return super()._rename(
5102 index, # type: ignore[arg-type]
5103 copy=copy,
5104 inplace=inplace,
5105 level=level,
5106 errors=errors,
5107 )
5108 else:
5109 return self._set_name(index, inplace=inplace, deep=copy)
File ~/work/pandas/pandas/pandas/core/generic.py:1140, in NDFrame._rename(self, mapper, index, columns, axis, copy, inplace, level, errors)
1138 return None
1139 else:
-> 1140 return result.__finalize__(self, method="rename")
File ~/work/pandas/pandas/pandas/core/generic.py:6281, in NDFrame.__finalize__(self, other, method, **kwargs)
6274 if other.attrs:
6275 # We want attrs propagation to have minimal performance
6276 # impact if attrs are not used; i.e. attrs is an empty dict.
6277 # One could make the deepcopy unconditionally, but a deepcopy
6278 # of an empty dict is 50x more expensive than the empty check.
6279 self.attrs = deepcopy(other.attrs)
-> 6281 self.flags.allows_duplicate_labels = other.flags.allows_duplicate_labels
6282 # For subclasses using _metadata.
6283 for name in set(self._metadata) & set(other._metadata):
File ~/work/pandas/pandas/pandas/core/flags.py:96, in Flags.allows_duplicate_labels(self, value)
94 if not value:
95 for ax in obj.axes:
---> 96 ax._maybe_check_unique()
98 self._allows_duplicate_labels = value
File ~/work/pandas/pandas/pandas/core/indexes/base.py:716, in Index._maybe_check_unique(self)
713 duplicates = self._format_duplicate_message()
714 msg += f"\n{duplicates}"
--> 716 raise DuplicateLabelError(msg)
DuplicateLabelError: Index has duplicates.
positions
label
b [0, 1]
Warning
This is an experimental feature. Currently, many methods fail to
propagate the allows_duplicate_labels value. In future versions
it is expected that every method taking or returning one or more
DataFrame or Series objects will propagate allows_duplicate_labels.