BUG: SeriesGroupBy.value_counts() raises on an empty Series

[FilippoBovo]

• I have searched the [pandas] tag on StackOverflow for similar questions.

• I have asked my usage related question on StackOverflow.

---

Question about pandas

Using the value_counts method after grouping an empty DataFrame and selecting a column raises an error.

Example:

>>> pd.DataFrame(columns=["A", "B"]).groupby("A")["B"].value_counts()

Error:

Traceback (most recent call last):
  File "<input>", line 1, in <module>
  File "/Users/username/.virtualenvs/my_project/lib/python3.8/site-packages/pandas/core/groupby/generic.py", line 736, in value_counts
    codes = [rep(level_codes) for level_codes in codes] + [llab(lab, inc)]
  File "/Users/username/.virtualenvs/my_project/lib/python3.8/site-packages/pandas/core/groupby/generic.py", line 705, in <lambda>
    llab = lambda lab, inc: lab[inc]
IndexError: boolean index did not match indexed array along dimension 0; dimension is 0 but corresponding boolean dimension is 1

Instead, I would expect that value_counts returns an empty series, like when using the max method, like the following,

>>> pd.DataFrame(columns=["A", "B"]).groupby("A")["B"].max()

which gives this result:

Series([], Name: B, dtype: object)

If it is normal to have an error when using value_counts for an empty SeriesGroupBy, why is it so?

Thank you.

[rhshadrach]

Thanks for the report. This looks like a bug to me. Further investigations and PRs to fix are welcome!

[AnnaDaglis]

Take

[AnnaDaglis]

That is the line that raises the error https://github.com/pandas-dev/pandas/blob/v1.2.1/pandas/core/groupby/generic.py#L736