How does networkx work?



In [1]:

    
# Import networkx and also matplotlib.pyplot for visualization
import networkx as nx
import matplotlib.pyplot as plt
%matplotlib inline

Undirected Graphs



In [12]:

    
# Create an empty undirected graph
G = nx.Graph()



In [13]:

    
# Add some nodes and edges.  Adding edges aslo adds nodes if they don't already exist.
G.add_node('Janos')
G.add_nodes_from(['Sophie', 'Otto'])
G.add_edge('Janos', 'Sophie')
G.add_edges_from([('Janos', 'Otto'), ('Sophie', 'Otto')])



In [14]:

    
nx.draw_spectral(G, with_labels=True, node_size=3000)
plt.show()

Directed Graphs



In [15]:

    
# Create an empty directed graph
G = nx.DiGraph()



In [16]:

    
# Add some nodes and edges.  Adding edges aslo adds nodes if they don't already exist.
G.add_node('Janos')
G.add_nodes_from(['Sophie', 'Otto'])
G.add_edge('Janos', 'Sophie')
G.add_edges_from([('Janos', 'Otto'), ('Sophie', 'Otto')])



In [17]:

    
nx.draw_spectral(G, with_labels=True, node_size=3000)
plt.show()

What can nodes be?

Nodes can be any hashable object: strings, integers, floats, functions, files, etc.

Both edges and nodes can store attributes in a dictionary.



In [18]:

    
import numpy as np
G.add_node(np.mean)
file = open('abc.txt', 'w')
G.add_node(file)
print(G.nodes())









    



[<_io.TextIOWrapper name='abc.txt' mode='w' encoding='cp1252'>, <function mean at 0x00000000054FFD90>, 'Janos', 'Otto', 'Sophie']

Reading in Data

We don't usually add data by hand.

NetworkX handles the following (and more) very well:

Standard files: txt, csv, json
Graph formats: GML, gexf
Python objects: dicts, Pandas DataFrames, lists of lists



In [23]:

    
!head addHealth81.txt









    



1	 463	1.00
1    679	1.00
1    958	1.00
1	1089	1.00
2     95	1.00
2    271	2.00
2    326	1.00
2    332	1.00
2    588	1.00
2    634	1.00



In [24]:

    
D = nx.read_weighted_edgelist('addHealth81.txt', create_using=nx.DiGraph())



In [25]:

    
len(D.nodes()), len(D.edges())









    Out[25]:





(12, 10)



In [ ]: