Helpful matplotlib snippets 📈

matplotlib -- the de-facto library of plotting in Python.

seaborn -- the golden tool for statistical data visualization. Also comes with nice style defaults.

I have found myself using both libraries frequently with some style preferences that I usually adopt when creating plots. For the sake of reproducibility and convenience, I have collected some of these snippets here.

Global style definitions

import matplotlib.pyplot as plt
import seaborn as sns

sns.set_theme(context='paper', style='ticks', font_scale=2)
plt.rcParams['font.family'] = 'Noto Sans'
plt.rcParams['axes.grid'] = True
plt.rcParams['axes.linewidth'] = 1.5

These give the plotted figure a clean outline with essential ticks on the axes.

Nicer bar plots

The default scheme in seaborn (on the left) is already quite nice, but with the following tweaks, we can make it even prettier.

Left: default seaborn scheme. Right: our tweaks.

We will be using this color palette.

palette = ["#4E79A7", "#F28E2B", "#E15759", "#59A14F", "#76B7B2", "#EDC949"]

import pandas as pd

# Mock some data here in a pandas DataFrame
data = pd.DataFrame(
    {
        "Fruit": ["Apple", "Banana", "Cherry"] * 4,
        "Quantity": [10, 15, 7, 12, 18, 5, 8, 11, 6, 11, 17, 4],
        "Storage": ["Fresh", "Dried", "Frozen", "Canned"] * 3,
    }
)

We plot the bar plot with seaborn on the axes we create with matplotlib. We apply a subtle transparency to the colors of the bars.

fig, ax = plt.subplots(figsize=(8, 8))

sns.barplot(
    data=data,
    x="Fruit",
    y="Quantity",
    hue="Storage",
    palette=palette,
    alpha=0.6,  # i've always found some transparency in bar plots to be helpful
    ax=ax,
)

Plot with transparency added.

Now, we are going to iterate over the bar patches within the plot, and:

• Set the edge color to a solid color with a thicker stroke.
• Add some breathing space between the bars.

gap = 0.2  # adjust this value to change the gap between bars

for _, bar in enumerate(ax.patches):  # may need to change this axis
    # Set edge color and linewidth
    clr = bar.get_facecolor()
    bar.set_edgecolor((*clr[:3], 1))  # opaque borders
    bar.set_linewidth(1.5)

    # Adjust the width and position of the bars
    width = bar.get_width()
    bar.set_width(width * (1 - gap))
    bar.set_x(bar.get_x() + (width - width * (1 - gap)) / 2)

Plot with edges and gaps added.

Finally, we will also add hatches to the bars for better visual distinction.

# Customize this list as preferred, options include:
# ['/', '-', '.', '/', '-', '.', '/', '-', '.', '/', '-', '.']
hatches = ['/', '-', '.']

for i, bar in enumerate(ax.patches):
    bar.set_hatch(hatches[i % len(hatches)])

Plot with hatches added.

For more hatch pattern options and examples, see Matplotlib Hatch Style Reference.

Saving figures

My only tip for saving figures is to apply .tight_layout() to the plot, and use bbox_inches='tight' to ensure that the saved figure does not have unnecessary whitespace around it.

fig.tight_layout()  # or plt.tight_layout()
plt.savefig('figure.png', dpi=300, bbox_inches='tight')

May be continued ...