Customizing Violin Plot Visualizations

Introduction

In this lab, we will learn how to customize violin plots using Matplotlib. Violin plots are a powerful tool for visualizing the distribution and density of data. By customizing the appearance of the plot, we can create more informative and visually appealing visualizations.

VM Tips

After the VM startup is done, click the top left corner to switch to the Notebook tab to access Jupyter Notebook for practice.

Sometimes, you may need to wait a few seconds for Jupyter Notebook to finish loading. The validation of operations cannot be automated because of limitations in Jupyter Notebook.

If you face issues during learning, feel free to ask Labby. Provide feedback after the session, and we will promptly resolve the problem for you.

Create test data

First, we will create some test data to use for the violin plot. We will use NumPy to generate four arrays of 100 normally distributed values with increasing standard deviations.

import matplotlib.pyplot as plt
import numpy as np

## create test data
np.random.seed(19680801)
data = [sorted(np.random.normal(0, std, 100)) for std in range(1, 5)]

Create a default violin plot

Next, we will create a default violin plot using Matplotlib's violinplot function. This will provide a baseline for comparison when we customize the plot in later steps.

## create default violin plot
fig, ax1 = plt.subplots()
ax1.set_title('Default Violin Plot')
ax1.set_ylabel('Observed Values')
ax1.violinplot(data)

Customize the violin plot appearance

Now we will customize the appearance of the violin plot. First, we will limit what Matplotlib draws by setting the showmeans, showmedians, and showextrema arguments to False. Then, we will change the color and opacity of the violin bodies using the set_facecolor and set_alpha methods. Finally, we will add a simplified representation of a box plot on top of the violin plot, using the percentile function from NumPy to calculate the quartiles, medians, and whiskers.

## customize violin plot appearance
fig, ax2 = plt.subplots()
ax2.set_title('Customized Violin Plot')
ax2.set_ylabel('Observed Values')

## create violin plot
parts = ax2.violinplot(
        data, showmeans=False, showmedians=False,
        showextrema=False)

## customize violin bodies
for pc in parts['bodies']:
    pc.set_facecolor('#D43F3A')
    pc.set_edgecolor('black')
    pc.set_alpha(1)

## add box plot
quartile1, medians, quartile3 = np.percentile(data, [25, 50, 75], axis=1)
whiskers = np.array([
    adjacent_values(sorted_array, q1, q3)
    for sorted_array, q1, q3 in zip(data, quartile1, quartile3)])
whiskers_min, whiskers_max = whiskers[:, 0], whiskers[:, 1]

inds = np.arange(1, len(medians) + 1)
ax2.scatter(inds, medians, marker='o', color='white', s=30, zorder=3)
ax2.vlines(inds, quartile1, quartile3, color='k', linestyle='-', lw=5)
ax2.vlines(inds, whiskers_min, whiskers_max, color='k', linestyle='-', lw=1)

Set axis style

Finally, we will set the style for the x-axis by specifying the tick labels and limits. We will define a helper function set_axis_style to accomplish this.

## set style for the axes
labels = ['A', 'B', 'C', 'D']
set_axis_style(ax2, labels)

def set_axis_style(ax, labels):
    ax.set_xticks(np.arange(1, len(labels) + 1))
    ax.set_xticklabels(labels)
    ax.set_xlim(0.25, len(labels) + 0.75)
    ax.set_xlabel('Sample Name')

Summary

In this lab, we learned how to customize the appearance of violin plots using Matplotlib. We created a default violin plot, then modified it by changing the color and opacity of the violin bodies, and adding a simplified box plot representation on top. We also set the style for the x-axis tick labels and limits. By customizing the appearance of the plot, we can create more informative and visually appealing visualizations.