Aggregation Unique supports the list format.

[mjshare]

Aggregation Unique supports the list format. Does the community have a plan?

[beckernick]

Could you provide a little bit more information?

Are you working in C++ or Python? Are you looking for a way to get the unique elements per list, the unique lists across a column, or something else?

[GregoryKimball]

Hello @mjshare thank you for your message. I believe the issue here is that we don't support groupby aggregations on List columns yet in cuDF-python.

We support unique on List columns:

>>> df = cudf.DataFrame({'a':[[1],[2],[1]], 'b':[1,2,3]})
>>> df['a'].unique()
0    [1]
1    [2]
Name: a, dtype: list

however, groupby throws a NotImplementedError

>>> df.groupby('a').min()
Traceback (most recent call last):
...
NotImplementedError: Unsupported column type passed to create an Index: <class 'cudf.core.column.lists.ListColumn'>

This issue appears to already be covered by #8039. For now I'll close this issue in favor of #8039 and the related #12037.