LOC methods doesn't change DATAFRAME's INDEX

[yutiansut]

Code Sample, a copy-pastable example if possible

Your code here

I use code in pandas guide as an example
http://pandas.pydata.org/pandas-docs/stable/generated/pandas.IndexSlice.html

while we using a loc to select some data we want, we hope the result has result's index, not the raw dataframe's index.

the example code in page could also say something:

here we make a multiindex_dataframe

import pandas as pd
import numpy as np
midx = pd.MultiIndex.from_product([['A0','A1'], ['B0','B1','B2','B3']])
columns = ['foo', 'bar']
dfmi = pd.DataFrame(np.arange(16).reshape((len(midx), len(columns))),
                        index=midx, columns=columns)

the we try to select some of them using loc

dfmi.loc[(slice(None), slice('B0', 'B1')), :]

we get 
Out[27]:
       foo  bar
A0 B0    0    1
   B1    2    3
A1 B0    8    9
   B1   10   11

this is all-right, however when we print the index of result

 dfmi.loc[(slice(None), slice('B0', 'B1')), :].index

we get

Out[28]:
MultiIndex(levels=[['A0', 'A1'], ['B0', 'B1', 'B2', 'B3']],
           labels=[[0, 0, 1, 1], [0, 1, 0, 1]])

this is the index of raw dataframe, now new dataframes' index

In financial, we using multiindex dataframe to contain data, so this problem maybe asked to be solved as soon as possible, thanks ~~

@yutiansut

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.6.5.final.0
python-bits: 64
OS: Windows
OS-release: 10
machine: AMD64
processor: Intel64 Family 6 Model 158 Stepping 10, GenuineInt
byteorder: little
LC_ALL: None
LANG: None
LOCALE: None.None

pandas: 0.23.0
pytest: 3.5.1
pip: 10.0.1
setuptools: 39.1.0
Cython: 0.28.2
numpy: 1.14.3
scipy: 1.1.0
pyarrow: 0.4.1
xarray: None
IPython: 6.4.0
sphinx: 1.7.4
patsy: 0.5.0
dateutil: 2.7.3
pytz: 2018.3
blosc: None
bottleneck: 1.2.1
tables: 3.4.3
numexpr: 2.6.5
feather: None
matplotlib: 2.2.2
openpyxl: 2.5.3
xlrd: 1.1.0
xlwt: 1.3.0
xlsxwriter: 1.0.4
lxml: 4.2.1
bs4: 4.6.0
html5lib: 1.0.1
sqlalchemy: 1.2.7
pymysql: 0.8.1
psycopg2: None
jinja2: 2.10
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None

[yutiansut]

@myles @jd @lebigot @hoffmann @parlarjb hope to get helped

[yutiansut]

currently the only thing to fix this is reset_indexand then set_index again

[jschendel]

This is the expected behavior. See the Defined Levels section of the docs. In particular, the remove_unused_levels method addresses this.

[yutiansut]

thanks a lot ~and a little suggestion @jschendel

maybe a link on IndexSlice page would be better ~

[jschendel]

Yes, this could definitely be documented better, as the behavior is somewhat surprising when it's first discovered. See #21308 for the related documentation issue I just created for this.