Python | Visualize graphs generated in NetworkX using Matplotlib

NetworkX is a Python library used to create and analyze graph structures. Although it's mainly for graph analysis, it also offers basic tools to visualize graphs using Matplotlib. In this article, you'll learn how to draw, label and save graphs using NetworkX's built-in drawing functions.

Steps to Visualize a Graph in NetworkX

1. Import the required libraries

import networkx as nx

import matplotlib.pyplot as plt

2. Generate a graph using networkx.Graph() or another graph class like DiGraph, MultiGraph, etc.

g.add_edge(1, 2)

3. Add edges (or nodes) to the graph using methods such as:

g.add_edge(1, 2)

4. Draw the graph using:

nx.draw(g)

5. (Optional) Add node labels using:

nx.draw(g, with_labels=True)

6. Save the plot as an image file using:

plt.savefig("filename.png")

Basic Example: Draw and Save a Graph

Output

Explanation: This code creates a simple undirected graph with 5 nodes and 5 edges, draws it and saves the output as filename.png.

Adding Node Labels

To include numbers (or labels) on nodes, simply set with_labels=True:

Output:

Explanation: Here, with_labels=True displays the node numbers in the graph, making it easier to understand connections.

Custom Layouts for Graph Drawing

NetworkX provides several graph layout algorithms to control how the graph is visually arranged. You can use various draw_* functions to apply different layouts:

Example with All Layouts

Circular Layout

Planar Layout

Random Layout

Spectral Layout

Spring Layout

Shell Layout

Explanation: A simple undirected graph is created using networkx.Graph() with added edges to define connections. The graph is visualized using six different NetworkX layouts, each saved as a separate image:

• nx.draw_circular() arranges nodes evenly in a circle, ideal for symmetrical or cyclic graphs.
• nx.draw_planar() tries to avoid edge crossings, working only for planar graphs.
• nx.draw_random() places nodes arbitrarily, offering quick but unstructured visuals.
• nx.draw_spectral() uses graph Laplacian eigenvalues to highlight structural patterns.
• nx.draw_spring() simulates nodes as repelling bodies with spring-like edges for a balanced look.
• nx.draw_shell() positions nodes in concentric circles, useful for layered or hierarchical data.

Related Articles

• Introduction to NetworkX Library
• Matplotlib
• Visualizing Geospatial Data
• Graph Theory with Python
• NetworkX Drawing Documentation