KeyError in express.scatter plot with facet_col when the data contains nan

[ryotatomioka]

Reproducer (modified from the original):

import plotly.express as px

df = px.data.tips()
df["sex"][: len(df) // 2] = None
fig = px.scatter(df, x="total_bill", y="tip", color="smoker", facet_col="sex")
fig.write_html("test.html")

The above produces the following error:

Traceback (most recent call last):
File "plotlyscatter.py", line 5, in
fig = px.scatter(df, x="total_bill", y="tip", color="smoker", facet_col="sex")
File "/opt/envs/myenv/lib/python3.8/site-packages/plotly/express/_chart_types.py", line 66, in scatter
return make_figure(args=locals(), constructor=go.Scatter)
File "/opt/envs/myenv/lib/python3.8/site-packages/plotly/express/_core.py", line 1976, in make_figure
group = grouped.get_group(group_name if len(group_name) > 1 else group_name[0])
File "/opt/envs/myenv/lib/python3.8/site-packages/pandas/core/groupby/groupby.py", line 754, in get_group
raise KeyError(name)
KeyError: ('No', '', '', '', nan)

Environment:

• plotly==5.3.1
• pandas==1.3.4

The simplest fix (which discards points with nan values) is to use grouped.indices instead of grouped.groups in this line:

plotly.py/packages/python/plotly/plotly/express/_core.py

Line 1914 in 5240301

for group_name in grouped.groups:

However it would be useful to allow nan values in grouping using dropna=False in DataFrame.groupby:
https://pandas.pydata.org/pandas-docs/dev/reference/api/pandas.DataFrame.groupby.html?highlight=groupby#pandas.DataFrame.groupby

[Tapot]

Got the same error with plotly 5.5.0, in 5.10.0 it is OK - looks like the problem was fixed

[pawel-kotowski]

I think I have similar issue in plotly 5.10.0 and 5.17.0 when there is no data in one of facet_row or facet_col cols when col is pd.Categorical. Works if I convert it to str type.