Create charts with seaborn

Use seaborn in executable code blocks to create statistical visualizations. Seaborn builds on matplotlib, so charts render as inline PNGs.

Setup #

Add seaborn and matplotlib to your site’s pyproject.toml:

[project]
dependencies = ["seaborn", "matplotlib"]

Then run uv sync in your site directory.

Distribution plot #

```{python}
import matplotlib
matplotlib.use('Agg')
import seaborn as sns
import matplotlib.pyplot as plt
import random

random.seed(42)
data = [random.gauss(0, 1) for _ in range(500)]

plt.figure(figsize=(8, 4))
sns.histplot(data, bins=30, kde=True, color='#7c3aed', alpha=0.7)
plt.title('Normal distribution (n=500)')
plt.tight_layout()
```

Here it is rendered:

import matplotlib
matplotlib.use('Agg')
import seaborn as sns
import matplotlib.pyplot as plt
import random

random.seed(42)
data = [random.gauss(0, 1) for _ in range(500)]

plt.figure(figsize=(8, 4))
sns.histplot(data, bins=30, kde=True, color='#7c3aed', alpha=0.7, edgecolor='white')
plt.title('Normal distribution (n=500)')
plt.xlabel('Value')
plt.ylabel('Frequency')
plt.grid(True, alpha=0.3, axis='y')
plt.tight_layout()

Box plot #

import matplotlib
matplotlib.use('Agg')
import seaborn as sns
import matplotlib.pyplot as plt
import random

random.seed(42)
data = {
    'Language': ['Rust'] * 50 + ['Python'] * 50 + ['Go'] * 50,
    'Build time (s)': (
        [random.gauss(2, 0.5) for _ in range(50)] +
        [random.gauss(5, 1.5) for _ in range(50)] +
        [random.gauss(3, 0.8) for _ in range(50)]
    ),
}

plt.figure(figsize=(8, 4))
sns.boxplot(x='Language', y='Build time (s)', data=data, palette=['#7c3aed', '#06b6d4', '#10b981'])
plt.title('Build times by language')
plt.grid(True, alpha=0.3, axis='y')
plt.tight_layout()

<string>:18: FutureWarning: 

Passing `palette` without assigning `hue` is deprecated and will be removed in v0.14.0. Assign the `x` variable to `hue` and set `legend=False` for the same effect.

Heatmap #

import matplotlib
matplotlib.use('Agg')
import seaborn as sns
import matplotlib.pyplot as plt
import random

random.seed(42)
matrix = [[random.random() for _ in range(6)] for _ in range(6)]
labels = ['A', 'B', 'C', 'D', 'E', 'F']

plt.figure(figsize=(7, 5))
sns.heatmap(matrix, annot=True, fmt='.2f', xticklabels=labels, yticklabels=labels, cmap='viridis')
plt.title('Correlation matrix')
plt.tight_layout()

How it works #

Seaborn creates matplotlib figures under the hood, so Zorto captures them exactly the same way — by detecting open figures via plt.get_fignums() and saving them as inline PNGs.

Use matplotlib.use('Agg') before importing seaborn to ensure the non-interactive backend is used during builds.