.loc with date string and timestamp behave differently

[KwangmooKoh]

Code Sample, a copy-pastable example if possible

>>> import pandas as pd
>>> t0 = pd.Timestamp('2010-01-01')
>>> t1 = pd.Timestamp('2012-02-02')
>>> df = pd.DataFrame([1, 2, 3], index=pd.MultiIndex.from_tuples([(t0, 'A'), (t0, 'B'), (t1, 'A')]), columns=['C'])

>>> # Compare the following results:
>>> print(df.ix['2010-01-01'])
#    C
# A  1
# B  2
>>> print(df.loc['2010-01-01'])
#               C
# 2010-01-01 A  1
#            B  2
>>> print(df.ix[t0])
#    C
# A  1
# B  2
>>> print(df.loc[t0])
#    C
# A  1
# B  2

Problem description

The result of df.loc['2010-01-01'] is different from that of df.ix['2010-01-01'] or df.loc[pd.Timestamp('2010-01-01')]; it contains additional index level for date. (df.ix[] returns the same data frame for date string and timestamp slicer.)

Expected Output

----
   C
A  1
B  2
----
   C
A  1
B  2
----
   C
A  1
B  2
----
   C
A  1
B  2
----

Output of pd.show_versions()

INSTALLED VERSIONS ------------------ commit: None python: 3.5.2.final.0 python-bits: 64 OS: Darwin OS-release: 15.6.0 machine: x86_64 processor: i386 byteorder: little LC_ALL: None LANG: en_US.UTF-8 LOCALE: en_US.UTF-8

pandas: 0.19.2+0.g825876c.dirty
nose: 1.3.7
pip: 8.1.2
setuptools: 27.2.0
Cython: 0.24.1
numpy: 1.11.2
scipy: 0.18.1
statsmodels: 0.6.1
xarray: None
IPython: 5.1.0
sphinx: 1.4.6
patsy: 0.4.1
dateutil: 2.5.3
pytz: 2016.6.1
blosc: None
bottleneck: 1.1.0
tables: 3.2.3.1
numexpr: 2.6.1
matplotlib: 1.5.3
openpyxl: 2.3.2
xlrd: 1.0.0
xlwt: 1.1.2
xlsxwriter: 0.9.3
lxml: 3.6.4
bs4: 4.5.1
html5lib: None
httplib2: None
apiclient: None
sqlalchemy: 1.0.13
pymysql: None
psycopg2: None
jinja2: 2.8
boto: 2.42.0
pandas_datareader: None

[jreback]

pls read the docs: http://pandas-docs.github.io/pandas-docs-travis/timeseries.html#indexing

this is all as expected

strings yield a slice, while a timestamp is a point
.ix is deprecated (and never did proper multi index partial string slicing)

[jorisvandenbossche]

@jreback Not sure is this is necessarily expected. It seems related to the discussion we had in #14826
cc @ischurov

Just looking a the .loc behaviour: the difference in return value seems to indicate that the string is interpreted as a slice, while the Timestamp as an exact match. But, the resolution of both the index as the string is 'day', so I think the string should be interpreted as an exact match as well?

Without the multi-index, both string as Timestamp return the same value (and are interpreted as an exact match):

In [2]: df2 = df.reset_index(level=1, drop=True)    

In [3]: df2
Out[3]: 
            C
2010-01-01  1
2010-01-01  2
2012-02-02  3

In [4]: df2.loc['2012-02-02']
Out[4]: 
C    3
Name: 2012-02-02 00:00:00, dtype: int64

In [5]: df2.loc[t1]
Out[5]: 
C    3
Name: 2012-02-02 00:00:00, dtype: int64

So at least there seems to be an inconsistency between index and multi-index?

[mroeschke]

Looks the code example in the OP is consistent now and I believe we have tests for this so closing