BUG: ensure_index inconsistently coerces np.nan to pd.NaT depending if timezones are present · Issue #27011 · pandas-dev/pandas (original) (raw)
Code Sample, a copy-pastable example if possible
On master:
In [1]: import numpy as np; import pandas as pd; pd.version Out[1]: '0.25.0.dev0+783.g2b9b58dad'
In [2]: ts_list = [pd.Timestamp('2018-01-01'), np.nan] ...: tstz_list = [pd.Timestamp('2018-01-01', tz='UTC'), np.nan]
In [3]: pd.core.indexes.base.ensure_index(ts_list) Out[3]: DatetimeIndex(['2018-01-01', 'NaT'], dtype='datetime64[ns]', freq=None)
In [4]: pd.core.indexes.base.ensure_index(tstz_list) Out[4]: Index([2018-01-01 00:00:00+00:00, nan], dtype='object')
Problem description
Out[4]
does not coerce np.nan
to pd.NaT
and results in an Index
with object
dtype instead of a DatetimeIndex
.
This causes downstream issues with IntervalIndex
/IntervalArray
as it can cause a valid IntervalIndex
/IntervalArray
to not be roundtripable from it's equivalent list
/np.array
representation:
In [5]: left = pd.DatetimeIndex(['2018-01-01', pd.NaT], tz='UTC') ...: right = pd.DatetimeIndex(['2018-01-02', pd.NaT], tz='UTC') ...: ii = pd.IntervalIndex.from_arrays(left, right) ...: ii Out[5]: IntervalIndex([(2018-01-01, 2018-01-02], nan], closed='right', dtype='interval[datetime64[ns, UTC]]')
In [6]: pd.IntervalIndex(ii.tolist())
TypeError: category, object, and string subtypes are not supported for IntervalIndex
In [7]: pd.IntervalIndex(ii.to_numpy())
TypeError: category, object, and string subtypes are not supported for IntervalIndex
Under the hood the list
/np.array
is being converted to left
/right
components, which are then passed to ensure_index
, resulting in an Index
with object
dtype, hence the error message.
Note that the equivalent roundtrip without a tz works fine, as expected based on the inconsistency noted in the ensure_index
example:
In [8]: left = pd.DatetimeIndex(['2018-01-01', pd.NaT]) ...: right = pd.DatetimeIndex(['2018-01-02', pd.NaT]) ...: ii = pd.IntervalIndex.from_arrays(left, right) ...: ii Out[8]: IntervalIndex([(2018-01-01, 2018-01-02], nan], closed='right', dtype='interval[datetime64[ns]]')
In [9]: pd.IntervalIndex(ii.tolist()) Out[9]: IntervalIndex([(2018-01-01, 2018-01-02], nan], closed='right', dtype='interval[datetime64[ns]]')
In [10]: pd.IntervalIndex(ii.to_numpy()) Out[10]: IntervalIndex([(2018-01-01, 2018-01-02], nan], closed='right', dtype='interval[datetime64[ns]]')
Expected Output
I'd expect Out[4]
to be coerced to a DatetimeIndex
with pd.NaT
and the appropriate tz:
Out[4]: DatetimeIndex(['2018-01-01 00:00:00+00:00', 'NaT'], dtype='datetime64[ns, UTC]', freq=None)
Output of pd.show_versions()
INSTALLED VERSIONS
commit : 2b9b58d
python : 3.7.3.final.0
python-bits : 64
OS : Linux
OS-release : 4.19.14-041914-generic
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : None
LANG : en_US.UTF-8
LOCALE : en_US.UTF-8
pandas : 0.25.0.dev0+783.g2b9b58dad
numpy : 1.16.4
pytz : 2019.1
dateutil : 2.8.0
pip : 19.1.1
setuptools : 40.8.0
Cython : 0.29.10
pytest : 4.6.2
hypothesis : 4.23.6
sphinx : 1.8.5
blosc : None
feather : None
xlsxwriter : 1.1.8
lxml.etree : 4.3.3
html5lib : 1.0.1
pymysql : None
psycopg2 : None
jinja2 : 2.10.1
IPython : 7.5.0
pandas_datareader: None
bs4 : 4.7.1
bottleneck : 1.2.1
fastparquet : 0.3.0
gcsfs : None
matplotlib : 3.1.0
numexpr : 2.6.9
openpyxl : 2.6.2
pandas_gbq : None
pyarrow : 0.11.1
pytables : None
s3fs : 0.2.1
scipy : 1.2.1
sqlalchemy : 1.3.4
tables : 3.5.2
xarray : 0.12.1
xlrd : 1.2.0
xlwt : 1.3.0
xlsxwriter : 1.1.8