CODEWITHSUNDEEP
pythonpandas
Carsten
Carsten

Reputation: 3048

Ghost indices in pandas MultiIndex

I have the following data:

arrays = [['bar', 'bar', 'baz', 'baz'],
           ['one', 'two', 'one', 'two']]

tuples = list(zip(*arrays))

index = pd.MultiIndex.from_tuples(tuples, names=['first', 'second'])

s = pd.Series(np.random.randn(4), index=index)

first  second
bar    one       1.791849
       two       0.334121
baz    one      -0.655277
       two      -1.296491


Now I filter on the index and remove "one":

t = s[s.index.get_level_values(1) != "one"]

first  second
bar    two       0.334121
baz    two      -1.296491


The strange thing is that "one" still shows up in the index (dependent on the method I use to access index values):

t.index.levels[1]

Index(['one', 'two'], dtype='object', name='second')

Or:

t.index.get_level_values(1)

Index(['two', 'two'], dtype='object', name='second')


Would you have any idea why "one" is still showing in the index?

Upvotes: 3

Views: 175

Answers (1)

jezrael
jezrael

Reputation: 862921

Use MultiIndex.remove_unused_levels, because by default after filtering is index not changed, I guess reason is performance:

print (t.index.remove_unused_levels())
MultiIndex([('bar', 'two'),
            ('baz', 'two')],
           names=['first', 'second'])

print (t.index.remove_unused_levels().levels[1])
Index(['two'], dtype='object', name='second')

Upvotes: 4

Related Questions