Skip to content

BUG: groupby with as_index=False drops groupings when names collide #58337

Open
@rhshadrach

Description

@rhshadrach

I think this is really and edge case because in order to get this behavior you need to group by an external Series with the same name as one of the columns.

df = pd.DataFrame({"a": [1, 1, 2], "b": [3, 4, 5]})

print(df.groupby(pd.Series([6, 6, 7], name="a"), as_index=False).sum())
#    a  b
# 0  2  7
# 1  2  5
print(df.groupby(pd.Series([6, 6, 7], name="a"), as_index=True).sum().reset_index())
# ValueError: cannot insert a, already exists

I think users expect as_index=False to behave the same as calling reset_index() after using as_index=True. It is not good behavior that we silently drop the groupings when as_index=False.

Metadata

Metadata

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions