Indexing a multiindex using a boolean series

December 24, 2021

I have a Series with a multiindex of companies and filing periods. I want to index this series with a separate boolean series of companies that pass some filter, returning all companies and filing periods that match the list of companies provided.

Here is example code:
First create a series with a multiindex.

arrays = [
    ["bar", "bar", "baz", "baz", "foo", "foo", "qux", "qux"],
    ["one", "two", "one", "two", "one", "two", "one", "two"],
]
tuples = list(zip(*arrays))
index = pd.MultiIndex.from_tuples(tuples, names=["first", "second"])
s = pd.Series(np.random.randn(8), index=index)

first  second
bar    one      -0.107011
       two       1.170427
baz    one      -0.567174
       two      -0.902847
foo    one      -0.986784
       two       0.972722
qux    one       0.714136
       two       0.044249
dtype: float64

Then create the boolean series (in my case generated from a groupby object with a filter for companies with multiple filing periods):

test = pd.Series([True, False, True, False], index=['bar', 'baz', 'foo', 'qux'], name='first')
bar     True
baz    False
foo     True
qux    False
Name: first, dtype: bool

I want to slice the Series to only return rows that match company bar and foo and omit the rest.

s.loc[test]
>>>IndexingError: Unalignable boolean Series provided as indexer (index of the boolean Series and of the indexed object do not match).

Can you help?

>Solution :

Try:

>>> s[test[test].index]
first  second
bar    one      -0.188867
       two       0.025214
foo    one       0.693238
       two       2.569597
dtype: float64