to_timedelta('1', unit='h') does not work.

[jtf621]

Code Sample, a copy-pastable example if possible

# Your code here

In [8]: pd.to_timedelta('1', unit='h')
Out[8]: Timedelta('0 days 00:00:00.000000')

Problem description

I would expect all the following to return Timedelta('0 days 01:00:00'). pd.to_timedelta(1, unit='h') does not. I must first cast the string '1' to a int or float then to a timedelta.

In [6]: pd.to_timedelta(1.0, unit='h')
Out[6]: Timedelta('0 days 01:00:00')

In [7]: pd.to_timedelta(1, unit='h')
Out[7]: Timedelta('0 days 01:00:00')

In [8]: pd.to_timedelta('1', unit='h')
Out[8]: Timedelta('0 days 00:00:00.000000')

In [9]: pd.to_timedelta('1 H')
Out[9]: Timedelta('0 days 01:00:00')

Expected Output

Output of pd.show_versions()

# Paste the output here pd.show_versions() here

INSTALLED VERSIONS

commit: None
python: 3.6.1.final.0
python-bits: 64
OS: Windows
OS-release: 10
machine: AMD64
processor: Intel64 Family 6 Model 78 Stepping 3, GenuineIntel
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: None.None

pandas: 0.20.2
pytest: 3.1.2
pip: 9.0.1
setuptools: 27.2.0
Cython: 0.25.2
numpy: 1.12.1
scipy: 0.19.0
xarray: None
IPython: 6.0.0
sphinx: None
patsy: 0.4.1
dateutil: 2.6.0
pytz: 2016.10
blosc: None
bottleneck: None
tables: 3.2.2
numexpr: 2.6.2
feather: None
matplotlib: 2.0.0
openpyxl: None
xlrd: 1.0.0
xlwt: None
xlsxwriter: None
lxml: None
bs4: 4.5.3
html5lib: 0.999999999
sqlalchemy: 1.1.5
pymysql: None
psycopg2: None
jinja2: 2.9.4
s3fs: None
pandas_gbq: None
pandas_datareader: None

[jorisvandenbossche]

IMO, we should raise an error in such a case (combination of specifying a unit and parsing a string). I recall we had a similar discussion for to_datetime.

If you have such a string, you should indeed first parse it to a numeric value (you can use to_numeric), and then to a timedelta. I personally don't think we should relax this requirement.

[jorisvandenbossche]

For to_datetime, issue is here: #15836 (and open PR: #15896), other related issue: #14350

[chris-b1]

I wouldn't have a problem raising, although it's worth noting that in to_datetime we do the "right" thing:

In [22]: pd.to_datetime(1, unit='D')
Out[22]: Timestamp('1970-01-02 00:00:00')

In [23]: pd.to_datetime('1', unit='D')
Out[23]: Timestamp('1970-01-02 00:00:00')

[jorisvandenbossche]

Hmm, OK, didn't check that :-)

As long as we raise on invalid ones, I am OK with that as well.
Eg the following works now, but should raise I think:

In [11]: pd.to_timedelta("1 min", unit='h')
Out[11]: Timedelta('0 days 00:01:00')

(similar case with to_timedelta does raise)

[jreback]

we need to raise if ANY unit is passed if its a string. The string contains or implies the unit.

A string is parsed against the timedelta spec e.g. .. days HH:MM:SS.FFFFFFFFF it will take partials so 1 is valid and since no unit is specified it IS ns by definition. e.g. equiv to 1 ns. Actually we should maybe disallow that as well (but in new issue).

[jorisvandenbossche]

Closing this as a duplicate of #12136