Skip to content

ENH: dtype-unaware (empty) objects ("any" dtype) #48110

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 11 commits into from
Closed

ENH: dtype-unaware (empty) objects ("any" dtype) #48110

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
11 commits
Select commit Hold shift + click to select a range
30e83ae
type
Aug 15, 2022
2cbc2e9
object
Aug 16, 2022
4a9a524
Merge branch 'main' into dtype-unaware-19647
Aug 16, 2022
d0c9c2b
add change
Aug 17, 2022
cf5a8e8
add object
Aug 17, 2022
387c237
Merge branch 'main' of https://github.com/weikhor/pandas into dtype-u…
Aug 18, 2022
22e7f10
Merge branch 'main' into dtype-unaware-19647
Aug 18, 2022
293c836
add test
Aug 18, 2022
ec5a8a2
Merge branch 'dtype-unaware-19647' of https://github.com/weikhor/pand…
Aug 18, 2022
647eff0
update test
Aug 18, 2022
a8a76a6
Merge branch 'main' into dtype-unaware-19647
Aug 18, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion pandas/core/indexing.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2117,7 +2117,8 @@ def _setitem_with_indexer_missing(self, indexer, value):
curr_dtype = getattr(curr_dtype, "numpy_dtype", curr_dtype)
new_dtype = maybe_promote(curr_dtype, value)[0]
else:
new_dtype = None
if isinstance(self.obj.dtype, object):
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't think this is a good idea. This breaks the current behavior and the special case for object looks a bit weird. What happens if you assign a different dtype? This would still change which is inconsistent then-

Also you have to keep the fall through clause

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

also wont this isinstance be True for everything?

Copy link
Contributor Author

@weikhor weikhor Aug 17, 2022 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

also wont this isinstance be True for everything? Yes. I should not do like this

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't think this is a good idea. This breaks the current behavior and the special case for object looks a bit weird. What happens if you assign a different dtype? This would still change which is inconsistent then-

Also you have to keep the fall through clause

From this issue #19647 , when adding element to empty object Series. The series should change based on type object which is set by user.

s = pd.Series(dtype='object')
s.loc['myint'] = 1
s.loc['myfloat'] = 2.

Expected output:

myint      1.0
myfloat    2.0
dtype: object

If I assign to different dtype, for example

s = pd.Series(dtype=float)
s.loc['myint'] = 1.0
s.loc['myfloat'] = 2

Output:

myint      1.0
myfloat    2.0
dtype: float64

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What happens if the empty series has some other dtype apart from object? Then this is not caught by your current change.

on the other side, if the dtype is the default this should still change, this is not covered here

Copy link
Contributor Author

@weikhor weikhor Aug 18, 2022 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What happens if the empty series has some other dtype apart from object? Then this is not caught by your current change.

If the empty series has some other dtype apart from object, for example

s = pd.Series(dtype=float)
s.loc['myint'] = 1.0
s.loc['myfloat'] = 2
print(s)

Output will be like this

myint      1.0
myfloat    2.0
dtype: float64

new_dtype = self.obj.dtype

new_values = Series([value], dtype=new_dtype)._values

Expand Down
15 changes: 15 additions & 0 deletions pandas/tests/series/indexing/test_setitem.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1665,3 +1665,18 @@ def test_setitem_empty_mask_dont_upcast_dt64():
ser.mask(mask, "foo", inplace=True)
assert ser.dtype == dti.dtype # no-op -> dont upcast
tm.assert_series_equal(ser, orig)


def test_setitem_on_series_dtype_object():
# GH#19647
result = Series(dtype="object")
result.loc["int"] = 1
result.loc["float"] = 2.0
expected = Series(data=[1, 2.0], index=["int", "float"]).astype("object")
tm.assert_series_equal(result, expected)

result = Series()
result.loc["int"] = 1
result.loc["float"] = 2.0
expected = Series(data=[1, 2.0], index=["int", "float"]).astype("float")
tm.assert_series_equal(result, expected)