The code below produces a very "dodgy" placement of the labels for edge weights in a graph. Please see image. I would like to have a better placement (close to the midpoint of each line), while still taking advantage of the automatic positioning of the nodes - i.e. I don't want to have to manually position the nodes.
Any ideas please? Also there is a warning - The iterable function was deprecated in Matplotlib 3.1 and will be removed in 3.3. Use np.iterable instead. which would be good to address if anyone knows how.
import matplotlib.pyplot as plt
import networkx as nx
import numpy as np
G = nx.Graph()
G.add_nodes_from(["A", "B", "C"])
G.add_edge("A", "B", weight=5)
G.add_edge("B", "C", weight=7)
G.add_edge("C", "A", weight=2)
pos = nx.spring_layout(G)
weights = nx.get_edge_attributes(G, "weight")
nx.draw_networkx(G, with_labels=True)
nx.draw_networkx_edge_labels(G, pos, edge_labels=weights)
plt.show()
解决方案draw_networkx(G, pos=None, arrows=True, with_labels=True, **kwds)
Parameters:
[...]
pos (dictionary, optional) – A dictionary with nodes as keys and positions as values. If not specified a spring layout positioning will
be computed. See networkx.layout for functions that compute node
positions.
So, if you do not pass pos explicitly, a spring_layout is generated, but this won't be identical to the layout that you generate through
pos = nx.spring_layout(G)
, because calling nx.spring_layout(G) twice gives different results:
for a in [0,1]:
pos = nx.spring_layout(G)
print(pos)
output:
{'A': array([ 0.65679786, -0.91414348]), 'B': array([0.34320214, 0.5814527 ]), 'C': array([-1. , 0.33269078])}
{'A': array([-0.85295569, -0.70179415]), 'B': array([ 0.58849111, -0.29820585]), 'C': array([0.26446458, 1. ])}
So, passing the same pos to both drawing functions solves the problem:
pos = nx.spring_layout(G)
weights = nx.get_edge_attributes(G, "weight")
nx.draw_networkx(G, pos, with_labels=True)
nx.draw_networkx_edge_labels(G, pos, edge_labels=weights)