How to index and slice a multiindex pandas dataframe

Question

In the below multiindex dataframe, how can I slice it so that I select rows where the index sign == 1 and the column csum_count == 8 ?

In [15]: raw3.head(20)
Out[15]:

                   csum_count  mean_return

sign cumsum

1    326              9        0.165
     854              9        0.081
     1346             9        0.055
     1440             9        0.157
     1554             9        0.069
     418              8        0.082
     578              8        0.119
     638              8        0.113
     896              8        0.076
     1480             8        0.059
     54               7        0.085
     446              7        0.120
     476              7        0.071
     484              7        0.094
     594              7        0.089
     622              7        0.069
     644              7        0.061
     1018             7        0.080
     1550             7        0.085
     1736             7        0.201

I'm able to select all rows with sign == 1 via the following: raw3.loc[(-1,)] . How do I add the additional criteria that csum_count == 8? . Thought that raw3.loc[(-1,),'csum_count'==8] might work but it didn't.

Answer 1

You can filter multi indices by calling DataFrame.index.get_level_values() and passing it the name (or level) of the index you want to slice on. In addition to that, you can add a condition on one of the columns.

raw3[(raw3.index.get_level_values('sign') == 1) & (raw3[csum_count == 8])]

Answer 2

You can use the.query() method for DataFrames, and pass a boolean string.

In your case, raw3.query('sign == 1 and cumsum == 8') should give you the result you want.