Floating point accuracy problems in Timestamp.round

[jeanmn]

This is the same problem as described in #14440 but with Timestamp, and the fix should be similar.

Probable source of error:

pandas/pandas/tslib.pyx

Line 438 in 093aa82

result = Timestamp(unit * rounder(value / float(unit)), unit='ns')

A small, complete example of the issue

In [1] import pandas as pd
In [2]: ts=pd.Timestamp('2016-03-06 18:43:45.950000')
In [3]: ts.floor('10L')
Out [3]: Timestamp('2016-03-06 18:43:45.950000128')  

Expected Output

Timestamp('2016-03-06 18:43:45.950000')

Output of pd.show_versions()

INSTALLED VERSIONS

commit: 093aa82
python: 2.7.6.final.0
python-bits: 64
OS: Linux
OS-release: 3.16.0-46-generic
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: None.None

pandas: 0.19.0+63.g093aa82
nose: 1.3.7
pip: 9.0.0
setuptools: 28.7.1
Cython: 0.25.1
numpy: 1.11.2
scipy: None
statsmodels: None
xarray: None
IPython: 5.1.0
sphinx: None
patsy: None
dateutil: 2.5.3
pytz: 2016.7
blosc: None
bottleneck: None
tables: None
numexpr: None
matplotlib: None
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml: None
bs4: None
html5lib: None
httplib2: None
apiclient: None
sqlalchemy: None
pymysql: None
psycopg2: None
jinja2: None
boto: None
pandas_datareader: None

[jreback]

duplicate

[jeanmn]

@jreback Issue link? I was on the verge of submitting a pull request to fix this

[jreback]

#14440 as u indicated

pr are welcome for this (it's an issue duplicate)

[jeanmn]

@jreback I see, thank you for the clarification.
I just wanted to make clear that this is a similar issue but at a different location in the code due to duplication of internal rounding logic. I wouldn't be surprised if the same error is found in more locations.
FYI: The current affected classes (so far) are Timestamp, Timedelta and DatetimeIndex