重复标签#
Index 对象不需要是唯一的；你可以有重复的行或列标签。这可能一开始有点令人困惑。如果你熟悉 SQL，你知道行标签类似于表上的主键，你永远不会希望在 SQL 表中有重复项。但 pandas 的角色之一是在数据进入下游系统之前清理混乱的现实世界数据。而现实世界的数据有重复项，即使在应该是唯一的字段中也是如此。
本节描述了重复标签如何改变某些操作的行为，以及如何在操作过程中防止重复标签的出现，或者在出现重复标签时如何检测它们。
In [1]: import pandas as pd
In [2]: import numpy as np
重复标签的后果#
一些 pandas 方法（例如 Series.reindex()）在存在重复项时无法工作。输出无法确定，因此 pandas 会抛出异常。
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 /home/pandas/pandas/core/series.py:4789, in Series.reindex(self, index, axis, method, copy, level, fill_value, limit, tolerance)
4772 @doc(
4773 NDFrame.reindex, # type: ignore[has-type]
4774 klass=_shared_doc_kwargs["klass"],
(...)
4787 tolerance=None,
4788 ) -> Series:
-> 4789 return super().reindex(
4790 index=index,
4791 method=method,
4792 level=level,
4793 fill_value=fill_value,
4794 limit=limit,
4795 tolerance=tolerance,
4796 copy=copy,
4797 )
File /home/pandas/pandas/core/generic.py:5347, in NDFrame.reindex(self, labels, index, columns, axis, method, copy, level, fill_value, limit, tolerance)
5344 return self._reindex_multi(axes, fill_value)
5346 # perform the reindex on the axes
-> 5347 return self._reindex_axes(
5348 axes, level, limit, tolerance, method, fill_value
5349 ).__finalize__(self, method="reindex")
File /home/pandas/pandas/core/generic.py:5369, in NDFrame._reindex_axes(self, axes, level, limit, tolerance, method, fill_value)
5366 continue
5368 ax = self._get_axis(a)
-> 5369 new_index, indexer = ax.reindex(
5370 labels, level=level, limit=limit, tolerance=tolerance, method=method
5371 )
5373 axis = self._get_axis_number(a)
5374 obj = obj._reindex_with_indexers(
5375 {axis: [new_index, indexer]},
5376 fill_value=fill_value,
5377 allow_dups=False,
5378 )
File /home/pandas/pandas/core/indexes/base.py:4191, in Index.reindex(self, target, method, level, limit, tolerance)
4188 raise ValueError("cannot handle a non-unique multi-index!")
4189 elif not self.is_unique:
4190 # GH#42568
-> 4191 raise ValueError("cannot reindex on an axis with duplicate labels")
4192 else:
4193 indexer, _ = self.get_indexer_non_unique(target)
ValueError: cannot reindex on an axis with duplicate labels
其他方法，如索引，可能会产生非常令人惊讶的结果。通常，使用标量进行索引会 降低维度 。使用标量对 DataFrame 进行切片将返回一个 Series 。使用标量对 Series 进行切片将返回一个标量。但在有重复项的情况下，情况并非如此。
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
我们在列中有重复项。如果我们切片 'B' ，我们会得到一个 Series
In [7]: df1["B"] # a series
Out[7]:
0 2
1 5
Name: B, dtype: int64
但是对 'A' 进行切片会返回一个 DataFrame
In [8]: df1["A"] # a DataFrame
Out[8]:
A A
0 0 1
1 3 4
这也适用于行标签
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
重复标签检测#
你可以检查一个 Index （存储行或列标签）是否唯一，使用 Index.is_unique 方法：
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
备注
检查索引是否唯一对于大型数据集来说是比较耗时的。pandas 会缓存这个结果，所以在同一个索引上重新检查会非常快。
Index.duplicated() 将返回一个布尔值的 ndarray，指示标签是否重复。
In [16]: df2.index.duplicated()
Out[16]: array([False, True, False])
这可以作为一个布尔过滤器来删除重复的行。
In [17]: df2.loc[~df2.index.duplicated(), :]
Out[17]:
A
a 0
b 2
如果你需要额外的逻辑来处理重复的标签，而不是仅仅丢弃重复项，使用 groupby() 对索引进行分组是一个常见的技巧。例如，我们将通过取具有相同标签的所有行的平均值来解决重复问题。
In [18]: df2.groupby(level=0).mean()
Out[18]:
A
a 0.5
b 2.0
不允许重复标签#
Added in version 1.2.0.
如上所述，在读取原始数据时处理重复项是一个重要的功能。也就是说，你可能希望避免在数据处理管道中引入重复项（通过 pandas.concat()、rename() 等方法）。Series 和 DataFrame 通过调用 .set_flags(allows_duplicate_labels=False) 来*禁止*重复标签（默认是允许的）。如果有重复标签，将引发异常。
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 /home/pandas/pandas/core/generic.py:464, in NDFrame.set_flags(self, copy, allows_duplicate_labels)
462 df = self.copy(deep=False)
463 if allows_duplicate_labels is not None:
--> 464 df.flags["allows_duplicate_labels"] = allows_duplicate_labels
465 return df
File /home/pandas/pandas/core/flags.py:118, in Flags.__setitem__(self, key, value)
116 if key not in self._keys:
117 raise ValueError(f"Unknown flag {key}. Must be one of {self._keys}")
--> 118 setattr(self, key, value)
File /home/pandas/pandas/core/flags.py:105, in Flags.allows_duplicate_labels(self, value)
103 if not value:
104 for ax in obj.axes:
--> 105 ax._maybe_check_unique()
107 self._allows_duplicate_labels = value
File /home/pandas/pandas/core/indexes/base.py:703, in Index._maybe_check_unique(self)
700 duplicates = self._format_duplicate_message()
701 msg += f"\n{duplicates}"
--> 703 raise DuplicateLabelError(msg)
DuplicateLabelError: Index has duplicates.
positions
label
b [1, 2]
这适用于 DataFrame 的行和列标签
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
这个属性可以通过 allows_duplicate_labels 进行检查或设置，该属性指示该对象是否可以具有重复标签。
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() 可以用来返回一个新的 DataFrame ，其属性如 allows_duplicate_labels 设置为某个值
In [24]: df2 = df.set_flags(allows_duplicate_labels=True)
In [25]: df2.flags.allows_duplicate_labels
Out[25]: True
返回的新 DataFrame 是与旧 DataFrame 相同数据的视图。或者属性可以直接在同一对象上设置。
In [26]: df2.flags.allows_duplicate_labels = False
In [27]: df2.flags.allows_duplicate_labels
Out[27]: False
在处理原始、混乱的数据时，你可能首先读入混乱的数据（可能包含重复标签），去重，然后禁止未来的重复，以确保你的数据管道不会引入重复项。
>>> raw = pd.read_csv("...")
>>> deduplicated = raw.groupby(level=0).first() # remove duplicates
>>> deduplicated.flags.allows_duplicate_labels = False # disallow going forward
在带有重复标签的 Series 或 DataFrame 上设置 allows_duplicate_labels=False 或在不允许重复的 Series 或 DataFrame 上执行引入重复标签的操作将引发 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 /home/pandas/pandas/core/frame.py:5583, in DataFrame.rename(self, mapper, index, columns, axis, copy, inplace, level, errors)
5461 """
5462 Rename columns or index labels.
5463
(...)
5580 4 3 6
5581 """
5582 self._check_copy_deprecation(copy)
-> 5583 return super()._rename(
5584 mapper=mapper,
5585 index=index,
5586 columns=columns,
5587 axis=axis,
5588 inplace=inplace,
5589 level=level,
5590 errors=errors,
5591 )
File /home/pandas/pandas/core/generic.py:1065, in NDFrame._rename(self, mapper, index, columns, axis, inplace, level, errors)
1063 return None
1064 else:
-> 1065 return result.__finalize__(self, method="rename")
File /home/pandas/pandas/core/generic.py:6030, in NDFrame.__finalize__(self, other, method, **kwargs)
6023 if other.attrs:
6024 # We want attrs propagation to have minimal performance
6025 # impact if attrs are not used; i.e. attrs is an empty dict.
6026 # One could make the deepcopy unconditionally, but a deepcopy
6027 # of an empty dict is 50x more expensive than the empty check.
6028 self.attrs = deepcopy(other.attrs)
-> 6030 self.flags.allows_duplicate_labels = other.flags.allows_duplicate_labels
6031 # For subclasses using _metadata.
6032 for name in set(self._metadata) & set(other._metadata):
File /home/pandas/pandas/core/flags.py:105, in Flags.allows_duplicate_labels(self, value)
103 if not value:
104 for ax in obj.axes:
--> 105 ax._maybe_check_unique()
107 self._allows_duplicate_labels = value
File /home/pandas/pandas/core/indexes/base.py:703, in Index._maybe_check_unique(self)
700 duplicates = self._format_duplicate_message()
701 msg += f"\n{duplicates}"
--> 703 raise DuplicateLabelError(msg)
DuplicateLabelError: Index has duplicates.
positions
label
X [0, 2]
Y [1, 3]
此错误消息包含重复的标签，以及所有重复项（包括“原始”）在 Series 或 DataFrame 中的数字位置。
重复标签传播#
一般来说，禁止重复是“粘性的”。它在操作中得以保留。
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 /home/pandas/pandas/core/series.py:4727, in Series.rename(self, index, axis, copy, inplace, level, errors)
4720 axis = self._get_axis_number(axis)
4722 if callable(index) or is_dict_like(index):
4723 # error: Argument 1 to "_rename" of "NDFrame" has incompatible
4724 # type "Union[Union[Mapping[Any, Hashable], Callable[[Any],
4725 # Hashable]], Hashable, None]"; expected "Union[Mapping[Any,
4726 # Hashable], Callable[[Any], Hashable], None]"
-> 4727 return super()._rename(
4728 index, # type: ignore[arg-type]
4729 inplace=inplace,
4730 level=level,
4731 errors=errors,
4732 )
4733 else:
4734 return self._set_name(index, inplace=inplace)
File /home/pandas/pandas/core/generic.py:1065, in NDFrame._rename(self, mapper, index, columns, axis, inplace, level, errors)
1063 return None
1064 else:
-> 1065 return result.__finalize__(self, method="rename")
File /home/pandas/pandas/core/generic.py:6030, in NDFrame.__finalize__(self, other, method, **kwargs)
6023 if other.attrs:
6024 # We want attrs propagation to have minimal performance
6025 # impact if attrs are not used; i.e. attrs is an empty dict.
6026 # One could make the deepcopy unconditionally, but a deepcopy
6027 # of an empty dict is 50x more expensive than the empty check.
6028 self.attrs = deepcopy(other.attrs)
-> 6030 self.flags.allows_duplicate_labels = other.flags.allows_duplicate_labels
6031 # For subclasses using _metadata.
6032 for name in set(self._metadata) & set(other._metadata):
File /home/pandas/pandas/core/flags.py:105, in Flags.allows_duplicate_labels(self, value)
103 if not value:
104 for ax in obj.axes:
--> 105 ax._maybe_check_unique()
107 self._allows_duplicate_labels = value
File /home/pandas/pandas/core/indexes/base.py:703, in Index._maybe_check_unique(self)
700 duplicates = self._format_duplicate_message()
701 msg += f"\n{duplicates}"
--> 703 raise DuplicateLabelError(msg)
DuplicateLabelError: Index has duplicates.
positions
label
b [0, 1]
警告
这是一个实验性功能。目前，许多方法未能传播 allows_duplicate_labels 值。在未来的版本中，预计每个接受或返回一个或多个 DataFrame 或 Series 对象的方法都将传播 allows_duplicate_labels。