TypeError with seaborn.kdeplot when using fill=True and categorical hue

mishachada commented 3 months ago

Description

I encountered a TypeError when using seaborn.kdeplot with the fill=True option and a categorical hue in my dataset. The error message indicates that there is a problem with data types being passed to the matplotlib fill function.

Steps to Reproduce

Here is a minimal reproducible example:

import matplotlib.pyplot as plt
import seaborn as sns
import pandas as pd
import numpy as np

# Generate synthetic datasets
np.random.seed(42)

data = pd.DataFrame({
    'value': np.concatenate([np.random.normal(5, 1, size=100), np.random.normal(10, 1, size=100), np.random.normal(15, 1, size=100)]),
    'category': ['Group 1']*100 + ['Group 2']*100 + ['Group 3']*100
})

# Plot using kdeplot
sns.kdeplot(data=data, x="value", hue="category", fill=True, palette='coolwarm', alpha=0.7)
plt.show()

Expected Behavior

The KDE plot should render successfully with the filled areas for different categories.

Actual Behavior

The plot does not render, and the following error is thrown: TypeError: ufunc 'isfinite' not supported for the input types, and the inputs could not be safely coerced to any supported types according to the casting rule ''safe''

Suggested Fix or Improvement

It seems that kdeplot with fill=True might need better handling for different data types or clearer documentation on the expected data types. Additionally, improving error messages to guide users to the correct data type could help prevent this issue.

A temporary workaround I found was using matplotlib directly for creating ridge plots, as shown below:

# Ridge plot workaround using matplotlib
from scipy.stats import gaussian_kde

fig, ax = plt.subplots()

for category in data['category'].unique():
    subset = data[data['category'] == category]['value']
    density = gaussian_kde(subset)
    xs = np.linspace(min(subset), max(subset), 200)
    ax.fill_between(xs, density(xs), alpha=0.6, label=category)

ax.legend()
plt.show()