熊猫数据框：如何定义几列属性？

Question

My question is so general that I'm pretty sure it might be a duplicate but I couldn't find the answer... For this, I apologize in advance !

I have the following issue :

Let's consider the pandas dataset obtained with the following lines:

import numpy as np
import pandas as pd

dates = pd.date_range( '20130101', periods=6 )
df = pd.DataFrame( np.random.randn( 6, 4 ), index=dates, columns=list( 'ABCD' ) )
df2 = pd.DataFrame ( [["dog", "dog", "snake", "cat"],["mammel", "mammel", "reptilian", "mammel"]], columns=list('ABCD'))
df2 = df2.append(df)

Which gives the following structure :

                            A         B          C          D
0                         dog       dog      snake        cat
1                      mammel    mammel  reptilian     mammel
2013-01-01 00:00:00  0.876641  -1.23665   0.375619  -0.642984
2013-01-02 00:00:00  -1.10135  0.264016  0.0893244     1.7381
2013-01-03 00:00:00 -0.551488  0.301711  -0.682017     1.4073
2013-01-04 00:00:00 -0.141452 -0.514909   -2.99863 -0.0283258
2013-01-05 00:00:00  -1.25555    1.1793   0.932212  -0.441398
2013-01-06 00:00:00  0.349649 -0.781957  0.0733772   0.810805

I would like to have the values corresponding to 0 and 1 as columns attributes. For instance I would like to keep all columns that correspond to mammel (A,B,D) but i'm just interested in values corresponding to dates.

So, in other words, I'm trying to find a way to do subsets of the original datasets through combinations of keywords or values (which I call attributes here).

I believe the same is possible in R dataframe structures but I could'nt find the best pythonic / pandatic way to do so !

Thanks in advance for any help provided !

Answer 1

Andrew's comment was exactly what I needed (and I owe you a big thanks for that) but I couldn't get the good words to find it myself.

The solution to my question was Multi-Indexing if someone ever reads this post !