stat_bin2d cuts values

[Hoeze]

Hi, I'd like to create a 2d bin plot of this dataset:

But as soon as I try to plot it:

plot = (
    pn.ggplot(df, pn.aes(x="score", y="residuals"))
    + pn.stat_bin_2d(binwidth=0.1)
)
plot

I just see this output:

However, I know that there should be points at score == 0:

plot = (
    pn.ggplot(df, pn.aes(x="score", y="residuals"))
    + pn.geom_point()
)
plot

What is the issue here?

[Hoeze]

Here is the dataset:
test.csv.gz

I am using plotnine v0.9.0

[Hoeze]

Hi @has2k1, is there something we can do about?
This is a really critical bug as it creates misleading results.

[has2k1]

Minimal Reprex

import pandas as pd
import numpy as np
from plotnine import *

n = 100
np.random.seed(123)
df = pd.DataFrame({
    'x': np.random.randint(0, 2, n),  # Issue is the 0
    'y': np.random.randn(n)
})

(ggplot(df, aes('x', 'y'))
 + stat_bin_2d(binwidth=0.1)
)

[Hoeze]

@has2k1 I'm sorry for bothering you, but is there any solution to this issue yet?

[has2k1]

I have a solution, I just wasn't satisfied with it. I will give it another look before the next couple of days and just publish it regardless.

[Hoeze]

@has2k1 Thanks for fixing this bug.
Would you mind also releasing a new plotnine version having the bug fix?