appending with multi-level series converts to single-level series

[abremod]

This was an issue that was fixed in 0.14.0, but has re-appeared in 0.18.1. The series s should have a mulit-level index, but in 18.1 it has a single-level index with a tuple for each value in the index.

Code Sample, a copy-pastable example if possible

import pandas as pd
from numpy.random import randn
a = [['bar', 'bar', 'foo', 'foo'],['one', 'two', 'one', 'two']]
t = list(zip(*a))
ind = pd.MultiIndex.from_tuples(t, names=['first', 'second'])
dat = pd.Series(randn(4), index=ind) 
s = pd.Series()
s = s.append(dat)
s.index.levels

Expected Output

FrozenList([['bar', 'foo'], ['one', 'two']])

output of pd.show_versions()

commit: None
python: 2.7.11.final.0
python-bits: 64
OS: Darwin
OS-release: 15.5.0
machine: x86_64
processor: i386
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8

pandas: 0.18.1
nose: 1.3.7
pip: 8.1.2
setuptools: 22.0.5
Cython: 0.24
numpy: 1.11.0
scipy: 0.17.1
statsmodels: 0.6.1
xarray: None
IPython: 4.2.0
sphinx: 1.4.1
patsy: 0.4.1
dateutil: 2.5.3
pytz: 2016.4
blosc: None
bottleneck: 1.0.0
tables: 3.2.2
numexpr: 2.5.2
matplotlib: 1.5.1
openpyxl: 2.3.2
xlrd: 1.0.0
xlwt: 1.1.1
xlsxwriter: 0.8.9
lxml: 3.6.0
bs4: 4.4.1
html5lib: None
httplib2: None
apiclient: None
sqlalchemy: 1.0.13
pymysql: None
psycopg2: None
jinja2: 2.8
boto: 2.40.0
pandas_datareader: None

[jreback]

So this is correct; the indexes of an empty Index does not align with a MultiIndex, so the multi-index is coerced to an Index (hence the tuples).

In [16]: Series(dat).append(dat)
Out[16]: 
first  second
bar    one       0.924763
       two       1.307207
foo    one      -0.811859
       two       0.381771
bar    one       0.924763
       two       1.307207
foo    one      -0.811859
       two       0.381771
dtype: float64

In [17]: Series().append(dat)
Out[17]: 
(bar, one)    0.924763
(bar, two)    1.307207
(foo, one)   -0.811859
(foo, two)    0.381771
dtype: float64

In [18]: Series().index
Out[18]: Index([], dtype='object')

I suppose this is a bit unexpected though and probably not the intent.

If you want to do a pull-request to conform it and if it doesn't break anything else, ok.

[jorisvandenbossche]

But it is correct that it is a regression from 0.18.0 to 0.18.1

[jreback]

ok possibly #12195, though more likely the various cleanups in Index w/empties are culprit.