python - 使用多个多索引级别删除

Question

我有一个像这样的多索引数据框:

      foo
c b a
p 6 1 3.0
q 7 2 2.3
r 8 3 1.0
s 9 4 100.0

我可以使用drop使用前 n MultiIndex 级别删除多行，如下所示:

>>> x.drop([('p', 6), ('r',8)])
      foo
c b a
q 7 2 2.3
s 9 4 100.0

我也可以drop从单一层面来看:

>>> x.drop([1, 2], level='a')
      foo
c b a
r 8 3 1.0
s 9 4 100.0

但我似乎无法对多个级别执行此操作(除了第一个 n):

>>> x.drop([(8, 3), (9, 4)], level=['b', 'a'])
Traceback (most recent call last):

  File "<ipython-input-156-a650ded10561>", line 1, in <module>
    x.drop([(8, 3), (9, 4)], level=['b', 'a'])

  File "/usr/lib/python2.7/dist-packages/pandas/core/generic.py", line 1399, in drop
    new_axis = axis.drop(labels, level=level)

  File "/usr/lib/python2.7/dist-packages/pandas/core/index.py", line 2718, in drop
    return self._drop_from_level(labels, level)

  File "/usr/lib/python2.7/dist-packages/pandas/core/index.py", line 2744, in _drop_from_level
    i = self._get_level_number(level)

  File "/usr/lib/python2.7/dist-packages/pandas/core/index.py", line 2199, in _get_level_number
    raise KeyError('Level %s not found' % str(level))

KeyError: "Level ['b', 'a'] not found"

这看起来很奇怪，因为 xs 确实接受级别列表，如示例所示:

>>> df.xs(('baz', 2), level=[0, 'third'])
        A  B  C  D
second
three   5  3  5  3

那么如何从数据框中删除 [(8, 3), (9, 4)] (即第三行和第四行)？

Answer 1

目前还没有此功能，请参阅此问题:https://github.com/pydata/pandas/pull/6599

但是您可以这样做。

In [19]: mask = df.index.get_level_values

In [20]: df.loc[~(mask('b').isin([8,9]) & mask('a').isin([3,4]))]
Out[20]: 
       foo
c b a     
p 6 1  3.0
q 7 2  2.3