sort_index with sort_remaining=False does not work with int level

[kdebrab]

Code Sample, a copy-pastable example if possible

import pandas as pd
mi = pd.MultiIndex.from_tuples([[2,2],[2,1],[1,2],[1,1]], names=list('AB'))
a = pd.Series([1,2,3,4], index=mi)
a.sort_index(level=0, sort_remaining=False)

returns:

A  B
1  1    4
   2    3
2  1    2
   2    1
dtype: int64

Problem description

Besides A (level 0), also B (level 1) is sorted. This, despite the fact that sort_remaining is set to False.

Expected Output

By comparison, the correct result is obtained when specifying the level by its name:

a.sort_index(level='A', sort_remaining=False)

with

A  B
1  2    3
   1    4
2  2    1
   1    2
dtype: int64

Exactly the same result should be expected when using level=0 instead of level='A'.

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 2.7.13.final.0
python-bits: 32
OS: Windows
OS-release: 10
machine: AMD64
processor: Intel64 Family 6 Model 78 Stepping 3, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None
LOCALE: None.None
pandas: 0.22.0
pytest: None
pip: 9.0.3
setuptools: 36.4.0
Cython: None
numpy: 1.14.1
scipy: 1.0.0
pyarrow: None
xarray: 0.10.0
IPython: 5.3.0
sphinx: 1.6.3
patsy: 0.5.0
dateutil: 2.6.1
pytz: 2018.3
blosc: None
bottleneck: 1.2.1
tables: None
numexpr: None
feather: None
matplotlib: 2.1.2
openpyxl: 2.5.0
xlrd: 1.0.0
xlwt: 1.3.0
xlsxwriter: 1.0.2
lxml: 4.1.1
bs4: 4.6.0
html5lib: 0.999999999
sqlalchemy: 1.2.3
pymysql: None
psycopg2: None
jinja2: 2.10
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None

[WillAyd]

Thanks for the bug. This is probably related to #21043, though the work done there does not appear to solve it.

Interestingly enough, the difference in sorting is ONLY observed at the first level. If your items were say at the second level there would be no difference.

In [2]: mi = pd.MultiIndex.from_tuples([[0,2,2],[0,2,1],[0,1,2],[0,1,1]], names=list('CAB'))
In [3]: a = pd.Series([1,2,3,4], index=mi)

In [5]: a.sort_index(level=1, sort_remaining=False)
Out[5]: 
C  A  B
0  1  2    3
      1    4
   2  2    1
      1    2
dtype: int64

In [6]: a.sort_index(level='A', sort_remaining=False)
Out[6]: 
C  A  B
0  1  2    3
      1    4
   2  2    1
      1    2
dtype: int64

[kdebrab]

Thus, the problem occurs only when level=0. Hmm, that sounds familiar. Indeed, a simple a.sort_index?? already reveals the culprit:

if level:
   [..]

In other words, level=0 is handled in the same way as level=None, which means that all levels are sorted...

The solution seems simple:

if level is not None:
   [..]

[WillAyd]

Perhaps - that's the same issue I found in the referenced PR #21043 though trying on that branch didn't resolve your issue. May be a similar issue somewhere else - investigation and PRs welcome!

[kdebrab]

I should have checked your link. I think you solved it for DataFrames, but not yet for Series...

[kdebrab]

Thanks for the quick response and fix @WillAyd !