I was a little surprised by this
In [11]: pd.Series([])
Out[11]: Series([], dtype: float64)
In [12]: pd.Series([None]) # should be NaN / float ?
Out[12]:
0 None
dtype: object
In [13]: pd.Series([None, 1])
Out[13]:
0 NaN
1 1
dtype: float64
Comment From: jreback
somewhat related to #7941. I think this should be Nan
(for a single None
) to be consistent.
Comment From: jorisvandenbossche
We recently discussed this in https://github.com/pandas-dev/pandas/pull/13894, where we sort of decided to not change this at the moment. So closing this, although we should maybe reconsider this somewhere for pandas 2.0 how to handle 'None'.