Plot nodes values in Networkx (Python)?

Question

I am importing a csv file called network_nodes which looks like

['151753', '111598', '0.211413517594337', '-0.130335792899132']
['151753', '118516', '0.211413517594337', '-0.100253812968731']

where the first two columns indicate nodes, and the 2 last columns are values associated with those nodes.

For instance, here the node called '151753' is connected to the node called '111598' and '118516'. And the node '151753' is associated with a value of '0.211413517594337', while '111598' is associated with a value of -0.130335792899132'.

I would like to plot this network in Networkx, using a different color (or node size) according to the node values (for instance red/big when the value is very high, blue/small when it is very low).

I do not know how to do this. I know that I should use something like

G=nx.read_adjlist('network_nodes.csv', delimiter=',')
nx.draw(G)

but the read_adjlist function do not allow me to import node values...

Answer 1

Aric is right that you need to write some custom code, but since you have a list of edges, and not a list of nodes, the following code will work better.

import csv
import networkx as nx

G = nx.DiGraph()
with open('network_nodes.csv') as f:
    node_list = csv.reader(f)
    for row in node_list:
        G.add_edge(row[0],row[1])
        G.node[row[0]]['value'] = float(row[2])
        G.node[row[1]]['value'] = float(row[3])

G.nodes(data=True) #print out

Prints the following:

[('151753', {'value': 0.211413517594337}),
 ('118516', {'value': -0.100253812968731}),
 ('111598', {'value': -0.130335792899132})]

Answer 2

You will need to write something custom to read your file. Here is one way:

node,weight,color
1,7.0,r
2,42,g
3,1,b

--

import csv
import networkx as nx
import matplotlib.pyplot as plt


G = nx.Graph()
with open('nodelist.txt') as f:
    reader = csv.DictReader(f)
    for row in reader:
        node = int(row.pop('node'))
        G.add_node(node, **row)

print G.nodes(data=True)
# [(1, {'color': 'r', 'weight': '7.0'}), (2, {'color': 'g', 'weight': '42'}), (3, {'color': 'b', 'weight': '1'})]

And drawing

nodes = G.nodes()
color = [G.node[n]['color'] for n in nodes]
size = [float(G.node[n]['weight'])*100 for n in nodes]
nx.draw(G, nodes=nodes,node_color=color, node_size=size)
plt.show()

Answer 3

The lines you quoted aren't in csv file format. Assuming you prefer having ints and floats instead of strings, this would be the data in correct csv format :

151753,111598,0.211413517594337,-0.130335792899132
151753,118516,0.211413517594337,-0.100253812968731

Either you can load your data as list, not using a csvreader, or you can convert your files with the following sed commands

sed -i.bak "s/\[\(.*\)\]$/\1/" network_nodes
sed -i.bak "s/'\([0-9\.-]*\)'/\1/g" network_nodes

.bak files are temporary backup files you can use to rollback a sed command. Remove them when you are done. Loading graph data with the csv loader should succeed after doing this.