CODEWITHSUNDEEP
pythonnetworkx
user8708316
user8708316

Reputation:

Make networkx plot look nice

I would need to build a nice network using the following data:

result_set = {('name1', 'job1'), ('name2', 'job2'), ('name3', 'job3'), ('name4', 'job4'), ('name5', 'job5'), ('name6', 'job6'), ('name7', 'job7'), ('name8', 'job8'), ('name9', 'job3'), ('name10', 'job6'), ('name11', 'job3'), ('name12', 'job1'), ('name13', 'job5'), ('name14', 'job9'), ('name15', 'job10'), ('name16', 'job6'), ('name17', 'job7'), ('name18', 'job11'), ('name19', 'job12'), ('name20', 'job13'), ('name21', 'job7'), ('name22', 'job14'), ('name23', 'job15'), ('name24', 'job7'), ('name25', 'job14'), ('name26', 'job9'), ('name27', 'job3'), ('name28', 'job16'), ('name29', 'job16'), ('name30', 'job1'), ('name31', 'job10'), ('name32', 'job9'), ('name33', 'job12'), ('name34', 'job5'), ('name35', 'job7'), ('name36', 'job3'), ('name37', 'job17'), ('name38', 'job3'), ('name39', 'job18'), ('name40', 'job16 / job3'), ('name41', 'Il Foglio'), ('name42', 'job7'), ('name43', 'job19'), ('name44', 'job9'), ('name45', 'job20'), ('name46', 'job18'), ('name47', 'job21')}

Names are unique, but not jobs, as you can see. I would like, therefore, to build a network that can show clusters of names by jobs.

I did it using the following code, but I am having some issues to visualise in a nice format, without overlapping the labels and changing nodes size depending on degree.

My code is the following:

result = zip(names, jobs)

# Converting itertor to set
result_set = set(result)
print(result_set)


G = nx.Graph()
for node_tuple in result_set:
        G.add_edges_from(result_set) # edited after comment
nx.draw(G, with_labels=True)
plt.show()

enter image description here

Could you please have a look at it and tell me how I could change it accordingly to let it better readable and show nodes depending on nodes' degree?

Suggestions and comments are always welcomed.

Upvotes: 6

Views: 6086

Answers (1)

yatu
yatu

Reputation: 88305

You can use one of the many layout algorithms implemented in networkx in nx.drawing.layout to position the nodes in a way that makes the visualisation of the network easier. You can further adjust the distance between the nodes by setting k to an appropriate value.

Also you can set the node size to be proportional to the degree by building a dict from Graph.degree and setting the node_size in nx.draw accordingly and scaling it up to the desired size by applying a multiplicative factor. Here's an example using spring_layout:

from pylab import rcParams
rcParams['figure.figsize'] = 14, 10
pos = nx.spring_layout(G, scale=20, k=3/np.sqrt(G.order()))
d = dict(G.degree)
nx.draw(G, pos, node_color='lightblue', 
        with_labels=True, 
        nodelist=d, 
        node_size=[d[k]*300 for k in d])

enter image description here

You could also custom the node colour according to whether a node is a job or a name using the node_color parameter (I'm guessing this is not the real case but it gives an idea of how to proceed):

rcParams['figure.figsize'] = 14, 10
pos = nx.spring_layout(G, scale=20, k=3/np.sqrt(G.order()))
colors = [['lightgrey', 'lightblue'][node.startswith('job')] 
          for node in G.nodes()]
d = dict(G.degree)
nx.draw(G, pos, 
        with_labels=True, 
        nodelist=d, 
        node_size=[d[k]*300 for k in d],
        node_color=colors)

enter image description here

Upvotes: 3

Related Questions