legend in python networkx

Question

I have the following code to draw a graph with nodes but i am failing to add a proper legend: (sorry, i can't post an image it seems i don't have enough reputation)

I want a legend with the 4 colors, such as "light blue = obsolese, red = Draft, Yellow = realease, dark blue = init".

I have seen some solutions with "scatter" but i think it is too complicated. Is there a way to do it with plt.legend(G.nodes)?

Here is the code:

import networkx as nx
import matplotlib.pyplot as plt
import numpy as np
G=nx.Graph()
G.add_node("kind1")
G.add_node("kind2")
G.add_node("Obsolete")
G.add_node("Draft")
G.add_node("Release")
G.add_node("Initialisation")
val_map = {'kind1': 2,'kind2': 2,'Obsolete': 2,'Initialisation': 1,'Draft': 4,'Release': 3}
values = [val_map.get(node, 0) for node in G.nodes()]
nodes = nx.draw(G, cmap = plt.get_cmap('jet'), node_color = values)

plt.legend(G.nodes())
plt.show()

score 8 · Answer 1 · edited Apr 11 '20 at 09:59

It seems that there is some kind of error when you are using nx.draw. Try to use nx.draw_networkx instead. And then use an axis from matplotlib to pass it when drawing the graph. This axis should contain the labels and colors of your nodes while plotting a point in (0,0) --> This is the tricky part.

Hope it helps! Here is the code I ran:

import networkx as nx
import matplotlib.pyplot as plt
import numpy as np
# For color mapping
import matplotlib.colors as colors
import matplotlib.cm as cmx

G=nx.Graph()
G.add_node("kind1")
G.add_node("kind2")
G.add_node("Obsolete")
G.add_node("Draft")
G.add_node("Release")
G.add_node("Initialisation")

# You were missing the position.
pos=nx.spring_layout(G)
val_map = {'kind1': 2, 
           'kind2': 2, 
           'Obsolete': 2, 
           'Initialisation': 1, 
           'Draft': 4, 
           'Release': 3}
values = [val_map.get(node, 0) for node in G.nodes()]
# Color mapping
jet = cm = plt.get_cmap('jet')
cNorm  = colors.Normalize(vmin=0, vmax=max(values))
scalarMap = cmx.ScalarMappable(norm=cNorm, cmap=jet)

# Using a figure to use it as a parameter when calling nx.draw_networkx
f = plt.figure(1)
ax = f.add_subplot(1,1,1)
for label in val_map:
    ax.plot([0],[0],
            color=scalarMap.to_rgba(val_map[label]),
            label=label)

# Just fixed the color map
nx.draw_networkx(G,pos, cmap=jet, vmin=0, vmax=max(values),
                 node_color=values,
                 with_labels=False, ax=ax)

# Here is were I get an error with your code                                                                                                                         
#nodes = nx.draw(G, cmap=plt.get_cmap('jet'), node_color=values)                                                                             

# Setting it to how it was looking before.                                                                                                              
plt.axis('off')
f.set_facecolor('w')

plt.legend(loc='center')

f.tight_layout()
plt.show()

Some useful sources:

I have an other question. Everytime i plot, the nodes' position change. But i want a constant position for the nodes ( i don't care what is the position of each nodes, i just want that if i plot severals times the graph, the nodes stay in the same position). Is there a paremeter in "pos=nx.spring_layout(G)" to add? I looked at http://networkx.github.io/documentation/latest/reference/generated/networkx.drawing.layout.spring_layout.html but i dont see a parameter that can be the solution. Thanks for your help!!! — Dupond, Apr 11 '14 at 09:28
Sure. I guess if you use a circular_layout instead of the spring_layout you will get something constant. You can check the layouts here: https://networkx.github.io/documentation/latest/reference/drawing.html — pequetrefe, Apr 12 '14 at 10:52

score 8 · Accepted Answer · edited May 08 '15 at 08:28

Thank you so much for your help, but it was not exactly what I wanted. I made few changes, so I can have the color legend's names differents from the nodes' names.

Here is the final code :

import networkx as nx
import matplotlib.pyplot as plt
import numpy as np
# For color mapping
import matplotlib.colors as colors
import matplotlib.cm as cmx

G=nx.Graph()
G.add_node("kind1")
G.add_node("kind2")
G.add_node("kind3")
G.add_node("kind4")
G.add_node("kind5")
G.add_node("kind6")

# You were missing the position.
pos=nx.spring_layout(G)
val_map = {'kind1': 2,'kind2': 2,'kind3': 2,'kind4': 1,'kind5':4,'kind6': 3}
#I had this list for the name corresponding t the color but different from the node name
ColorLegend = {'Obsolete': 2,'Initialisation': 1,'Draft': 4,'Release': 3}
values = [val_map.get(node, 0) for node in G.nodes()]
# Color mapping
jet = cm = plt.get_cmap('jet')
cNorm  = colors.Normalize(vmin=0, vmax=max(values))
scalarMap = cmx.ScalarMappable(norm=cNorm, cmap=jet)

# Using a figure to use it as a parameter when calling nx.draw_networkx
f = plt.figure(1)
ax = f.add_subplot(1,1,1)
for label in ColorLegend:
    ax.plot([0],[0],color=scalarMap.to_rgba(ColorLegend[label]),label=label)

# Just fixed the color map
nx.draw_networkx(G,pos, cmap = jet, vmin=0, vmax= max(values),node_color=values,with_labels=True,ax=ax)

# Setting it to how it was looking before.                                                                                                              
plt.axis('off')
f.set_facecolor('w')

plt.legend()

f.tight_layout()
plt.show()

score 1 · Answer 3 · answered Nov 13 '22 at 18:47

I tried different approach which worked for my case.

Here is the code i used

from matplotlib.lines import Line2D

legend_elements = [
    Line2D([0], [0], marker='o', color='w', label='Label1',markerfacecolor='g', markersize=15),
    Line2D([0], [0], marker='o', color='w', label='label2',markerfacecolor='r', markersize=15),        
]

nx.draw_networkx(G, pos=nx.spring_layout(G),edge_color=(0.8,0.6,0.3), node_color=color)

plt.legend(handles=legend_elements, loc='upper right')
plt.savefig('network_graph.png')

legend in python networkx

3 Answers3