[Python] polars groupby on categorical produces nulls

Update: This issue is no longer present. The query produces the expected values.


I ran into an issue with polars where after a groupby on a categorical variable the labels are null

import polars as pl

scores = pl.DataFrame({
'zone': ['North', 'North', 'South', 'South'],
'score': [78, 39, 76, 56]
}).with_columns(pl.col("zone").cast(pl.Categorical))

scores.group_by("zone").len()


This produces the output below with null values for the levels. Is this expected? If so, is there a fix? Note that this doesn't happen when zone is set to a string

shape: (2, 2)
┌──────┬─────┐
│ zone ┆ len │
│ --- ┆ --- │
│ cat ┆ u32 │
╞══════╪═════╡
│ null ┆ 2 │
│ null ┆ 2 │
└──────┴─────┘

Continue reading...