statsmodel raises an error when plotly tries to fit a trendline to datetime-vs-datetime scatterplot

[gvwilson]

1. Create a dataframe with two columns sample_date and analysis_date, both of type datetime64. Neither column has missing values.
2. Create a scatterplot with px.scatter(df, x="sample_date", y="analysis_date") - works fine.
3. Create another scatterplot with a trendline using px.scatter(df, x="sample_date", y="analysis_date", trendline="ols"): blows up.

The error message is:

numpy.core._exceptions._UFuncBinaryResolutionError: ufunc 'subtract' cannot use operands with types dtype('<M8[ns]') and dtype('O')

A workaround is to convert the datetime64 columns to ordinal numbers:

for colname in ("sample_date", "analysis_date"):
    df[colname] = df[colname].map(datetime.toordinal)

but then the ticks are wrong (large integers instead of datetimes). This can be partially corrected by specifying the ticks:

tick_x = df['sample_date'].map(lambda x: datetime.fromordinal(x).strftime('%b %d'))
fig.update_xaxes(tickvals=df['sample_date'], ticktext=tick_x)
tick_y = df['analysis_date'].map(lambda x: datetime.fromordinal(x).strftime('%b %d'))
fig.update_yaxes(tickvals=df['analysis_date'], ticktext=tick_y)

but then the spacing of ticks/lines along the axes is uneven (see screenshot).

Desired behavior: trendlines with time-series data should work out of the box.

[gvwilson]

Hi - we are trying to tidy up the stale issues and PRs in Plotly's public repositories so that we can focus on things that are still important to our community. Since this one has been sitting for several years, I'm going to close it; if it is still a concern, please add a comment letting us know what recent version of our software you've checked it with so that I can reopen it and add it to our backlog. Thanks for your help - @gvwilson