Regression line not appearing over scatterplot when using .add(so.Line(), so.PolyFit())

[joshua-ebner]

I'm trying to create a scatterplot with a regression line over the top. When I try to do this with .add(so.Line(), so.PolyFit()) the trend line fails to appear.

Here is the code that I'm using to attempt to create the scatterplot + regression line:

import pandas as pd
import seaborn.objects as so

supercars = pd.read_csv('https://learn.sharpsightlabs.com/datasets/pdm/supercars.csv')


(so.Plot(data = supercars
         ,x = 'horsepower'
         ,y = 'top_speed'
         )
   .add(so.Dot())
   .add(so.Line(), so.PolyFit())
 )

And here is the output:

As you can see, the scatterplot plots as expected, but the regression line is missing.

To show you what I'm expecting to appear, I'll create a "regplot" with 'sns.regplot':

import seaborn as sns

sns.regplot(data = supercars
         ,x = 'horsepower'
         ,y = 'top_speed'
         )

OUT:

As you can see, with the traditional "regplot" the scatterplot has a regression line over it.

I think that in my Seaborn Objects code is correct, but perhaps I've made a mistake.

If not, and my code is correct, then it seems that there's a bug in how Seaborn Objects is operating on this particular dataset.

[mwaskom]

Are there missing values in your dataset? cf #3010

[joshua-ebner]

Yes.

One of the two variables, top_speed, has 10 missing values.

supercars.horsepower.isna().sum() #0 records
supercars.top_speed.isna().sum() #10 records

To be clear: I was unaware that this would be an issue, since there wasn't an error message, and since it worked with sns.regplot in the past.

[jcmkk3]

I think that what @mwaskom meant was that this is a known issue and has been addressed. If you upgrade to the latest version (released in the last few days), the fix should be included.

As an aside, python allows trailing commas for the final item so there's no advantage to writing your function calls with leading commas.

This:

 sns.regplot(data = supercars
          ,x = 'horsepower'
          ,y = 'top_speed'
          )

Can be written like this:

 sns.regplot(
     data=supercars,
     x='horsepower',
     y='top_speed',
 )

Of course, you're welcome to style your code any way that you'd like.

[mwaskom]

Yep then this is fixed (also I’m pretty sure there was warning message about nonfinite data or something coming out of numpy that provided some clues).