Network Graph Visualization

Introduction

This package creates a visualization of a network graph built with Networkx with hovering functions by Plotly.

Multiple node and edge attributes can be added to the network and shown in the visualization.

For the full package description go to https://test.pypi.org/project/network-graph-visualization/.

And for the documentation go to the corresponding Homepage at https://network-graph-visualization.readthedocs.io/en/latest/

Prerequisites

This package requires networkx version >= 2.5 and plotly version >= 4.14.3. See pyproject.toml for the complete prerequisites.

Installation

Start with install the dependencies: pandas, matplotlib, networkx, plotly, Sphinx.

Install the package through:

pip install -i https://test.pypi.org/simple/ network-graph-visualization

Usage

Below is an example how to use this package. This description also shows how to add node and edge attributes to the graph from the corresponding pandas dataframes. The output will be anything similar to the below figures.

figures/Node_Attributes.png figures/Edge_Attributes.png

Create two separate dataframes. One with information about the nodes and one with information about the connections. For simplicity, call them connections_df and nodes_df.

connections_df = pd.read_csv(CONNECTIONS_FILENAME)
nodes_df = pd.read_csv(NODES_FILENAME)

Build an empty graph.

G = nx.Graph()

Add edge attributes. Create a column of connections as input for Networkx. Set these as the index, Convert dataframe to dictionary where the indices are the key and the attributes the values. Add edges and their attributes to empty graph.

connections_df['connections'] = list(zip(connections_df['SOURCE_VARIABLE'], connections_df['TARGET_VARIABLE']))
connections_temp = connections_df[['connections', 'EDGE_ATTRIBUTE_1', 'EDGE_ATTRIBUTE_2']].set_index('connections')
connections_dict = connections_temp.to_dict('index')
G.add_edges_from((k[0], k[1], d) for k,d in connections_dict.items())

Add node attributes. In contrast to edge attributes node attributes can be added all at once.

nodes_temp = nodes_df.set_index('NODE_NAME_VARIABLE')
nodes_dict = nodes_temp.to_dict('index')
nx.set_node_attributes(G, nodes_dict)

Call the package.

import network_graph_visualization.plot
network_plot = network_graph_visualization.plot.GraphNetwork(G)

View graph attributes.

print(network_plot.G.nodes(data=True))
print(network_plot.G.edges(data=True))

Optional to add all node and edge attributes as hovering text.

Add node hover text.

NODE_HOVERTEXT = []
for node in G.nodes():
    NODE_HOVERTEXT.append(
                    "Name: " + node + "<br>" + \
                    "NODE_ATTRIBUTE_1: " + str(network_plot.G.nodes[node]['NODE_ATTRIBUTE_1']) + "<br>" + \
                    "NODE_ATTRIBUTE_2: " + str(network_plot.G.nodes[node]['NODE_ATTRIBUTE_2'])
                    )

Add edge hover text.

EDGE_HOVERTEXT = []
for edge in G.edges():
    EDGE_HOVERTEXT.append(
                    "EDGE_ATTRIBUTE_1: " + str(G.edges[edge]['EDGE_ATTRIBUTE_1']) + "<br>" + \
                    "EDGE_ATTRIBUTE_2: " + str(G.edges[edge]['EDGE_ATTRIBUTE_2'])
                    )

Run node and edge traces.

network_plot.trace_nodes(node_color_variable='NODE_ATTRIBUTE_1', node_text=NODE_HOVERTEXT)
network_plot.trace_edges(edge_text=EDGE_HOVERTEXT) #edge_attribute='EDGE_ATTRIBUTE_2'

Build visualization.

network_plot.visualization_attributes(title='TITLE OF THE PLOT')

Draw and visualize the network.

network_plot.draw_network(graph_filename='GRAPH_FILENAME.html')

Authors and acknowledgment

Annalie Kruseman

Feel free to contact me about any questions related to this package. annaliakruseman@gmail.com

This package has been tested with the Stack Overflow Tag Network as posted on Kaggle.

Documentation

class network_graph_visualization.plot.GraphNetwork(G)

An object that builds a network graph in Plotly with Networkx including functions that display node and edge attributes.

Object requires a graph built with Networkx. If node and edge attributes are to be added to the visualization those attributes need to be added to the graph.

draw_network(graph_filename='graph.html')

Draw the network.

Parameters

graph_filename (str, default='graph.html') – Name and location of the filename as ‘html’.

Returns

Output file with ‘.html’ extension.

trace_edges(edge_text=None, edge_attribute=None)

Create a function to build a dictionary to trace edges, including the text when hovering over an edge.

Parameters

  • edge_text (list of strings, default=None) – The hover text corresponding to each edge. The list of strings consists of multiple edge attributes each attribute separated by <br> to indicate a new line. The list should be ordered by the order in which the edges appear as in G.edges().

  • edge_attribute (str, default=None) – The name of the variable to show as a single edge attribute when hovering over the edge. When edge_text equals None AND edge_attribute equals None, no edge attributes are shown.

trace_nodes(node_color_variable=None, node_text=None)

Create a function to build a dictionary to trace nodes, including the color and text when hovering over an edge.

Parameters

  • node_color_variable (str) – The name of the variable that distinguishes nodes by color. The values of the colors are represented as integers. Default equals None when no color is assigned to the nodes.

  • node_text (list of strings) – The hover text corresponding to each node. The list of strings consists of multiple node attributes each attribute separated by <br> to indicate a new line. The list should be ordered by the order in which the nodes appear as in G.nodes(). The default node text equals None, which means that the name of the node is used when hovering over the node.

visualization_attributes(title=<class 'str'>)

Set axis attributes.

Default to hiding axis lines, hiding grid, hiding ticklabels.

Parameters

  • title (str) – Title of the graph network plot.

  • showline (bool) – Defaults to False

  • zeroline (bool) – Defaults to False

  • showgrid (bool) – Defaults to False

  • showticklabels (bool) – Defaults to False

  • width (int) – Width of the plot.

  • height (int) – Height of the plot.

  • autosize (bool) – Defaults to False

  • showlegend (bool) – Defaults to False

  • margin (dictionary of key : int value pairs) – l=40,r=40,b=85,t=100,pad=0

  • hovermode ({'closest'}) – When hovermode equals ‘closest’ a single hover label appears for the point directly underneath the cursor. Defaults to ‘closest’.

  • plot_bgcolor (str) – Set background color. Defaults to ‘#ffffff’

Indices and tables