Selecting one repeated multiindex in a dataframe on pandas

Question

major minor col
0     0     5
      1     6
      2     4
0     0     8
      1     5
      2     6
1     0     3
      1     6
      2     9
1     0     5
      1     1
      2     7

First I'd like to get

major minor col
0     0     5
      1     6
      2     4
0     0     8
      1     5
      2     6

and then select over both major '0's, that is, choose the first major 0 or the second:

major minor col
0     0     5
      1     6
      2     4

or

major minor col
0     0     8
      1     5
      2     6

Unfortunately df.xs(0,level=0,drop_level=False) doesn't exactly fit the job, since it maintains major '1's in the index, although empty. Any ideas?

Answer 1

I still do not understand your data structure. I'm right now working with

                 val
major minor col     
0     0     5      1
      1     6      1
      2     4      1
      0     8      1
      1     5      1
      2     6      1

I still don't understand how in your case you have two major zeros, since I get only one with the same structure. Therefore, I can't tell you exactly how you could pick any of the *major*s.

Using traditional slicing, you can get where df.major == 0 using

df[df.major == 0]

In order to select any single of the subgroups now, it depends on how they're different. Do they have another unique feature? Then you could do

df[(df.major == 0) && (df.someColumn == someValue)]

(notice the brackets). Otherwise, if you know there are 3 rows per group, df[df.major == 0].iloc[:3] (or 3: ) will give you the records.

Also, have a look at the (currently experimental) df.query() ( documentation ).

Generally, you can do stuff such as

df[df.major == 0] to get all the values where the major is zero. If it's a (labeled) index or a normal column does not matter. You can also stack these to do

`df[(df.major == 0)

I start with

In[264]: df
Out[262]: 
                 val
major minor col     
0     0     5      1
      1     6      1
      2     4      1
      0     8      1
      1     5      1
      2     6      1
1     1     3      1
            6      1
      2     9      1
      1     5      1
            1      1
      2     7      1

and then I do

In[263]: df.query('major == 0')
Out[261]: 
                 val
major minor col     
0     0     5      1
      1     6      1
      2     4      1
      0     8      1
      1     5      1
      2     6      1