MultiIndexing Issue

[Aurthes]

Let 'b' be a large dataframe with MulitiIndex.

I pull the 4th block based on the 1st level of the index, and I got the block fine.

a = b.ix[b.index.levels[0][4]]

Then I printed the index of the block, which also seemed fine, code and result as follows:

print a.index

sequence attribute
4 count
price
quantity

and when I pull values, they seemed fine too:

a.index.values

array([(4L, 'count'), (4L, 'price'), (4L, 'quantity')], dtype=object)

but when I looked at it in a different way, things didn't make much sense:

a.index

MultiIndex(levels=[[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, ...], [u'count', u'price', u'quantity']],
labels=[[4, 4, 4], [0, 1, 2]],
names=[u'sequence', u'attribute'])

and when I pull the first level from the index of the block, I got:

a.index.levels[0]

Int64Index([0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, ...], dtype='int64')

Which is not the

Int64Index([4], dtype='int64')

I anticipated.

Of course I can

pd.MultiIndex.from_tuples(a.index.values)

and create the thing I am expecting, but this is a pain in the ass. Any solutions? Thanks!

[jreback]

you will have to show some copy-pastable example. not really sure what you are trying to do, nor what error you are seeing. and show pd.show_versions()

[Aurthes]

import pandas as pd
pd.show_versions()

def block_factory(lvl1,lvl2):
    idx1 = [lvl1]
    idx2 = [lvl2]
    idx3 = ['ha','haha','hahaha']
    idx = pd.MultiIndex.from_product([idx1,idx2,idx3])
    df = pd.DataFrame(range(3),index=idx)
    return df
df1=block_factory(1,'a')
df2=block_factory(2,'b')
df = pd.concat((df1,df2))
df_temp = df.ix[df.index.levels[0][1]]
df_temp
df_temp.index.levels[0]

I am anticipating 'b' only, while I get ['a','b']

[jreback]

I want you to pd.show_versions() and post it

[jreback]

using .levels is a semi-private method.
The reason that the levels are still their (but not indicated by the labels) is an efficiency/impl detail. You can reconstruct the index if you want, but it really doesn't affect anything. A more full description is in this issue #2770

In [16]: df_temp.index.get_level_values(0)
Out[16]: Index([u'b', u'b', u'b'], dtype='object')

In [17]: df_temp.index.get_level_values(1)
Out[17]: Index([u'ha', u'haha', u'hahaha'], dtype='object')