networkx 社交网络分析：绘制子图

如何解决networkx 社交网络分析：绘制子图

使用 networkx，我有一个包含 20 个具有 eignvector 中心性的节点的列表。我想制作一个社交网络分析的子图，其中边和节点连接到这些前 20 个节点。

sorted_eigenvector = sorted(eigenvector_dict.items(),key=itemgetter(1),reverse=True)

print("Top 20 nodes by eigenvector centrality:")
for b in degree[:20]:
    print(b)

返回

特征向量中心性的前 20 个节点： ('shafiur',0.45395678017648816) ('JusticeMyanmar',0.3901833956501838) ('drzarni',0.36392376672758797) ('Reaproy',0.2588852510372045) ('maksimbenenson',0.23262562211460833) ('SAdamsR2P',0.1951118665108976) ('uscb',0.19136049095052945) ('meemeeyeemon',0.18760911539015862) （'本尼迪克特人'，0.18010636426941695） ('KoSoe_T',0.17635498870904612) ('forum_asia',0.17635498870904612) ('hninyadanazaw',0.1763549887090444) ('MayWongCNA',0.16134948646756175) ('UNCANews',0.14634398422607858) ('akilaGJC',0.14634398422607858) ('loucharbon',0.14634398422607858) ('GCR2P',0.13884123310533691) ('mahninpwint',0.10507885306200064) ('LeongWaiKitCNA',0.09007335082051672) ('Milktea_Myanmar',0.08257059969977519)

H = G.subgraph('shafiur')
H_neighbors = ["shafiur"]

for edge in G.edges():
  
  if edge[0]=="shafiur":
    H_neighbors.append(edge[1])


  elif edge[1]=="shafiur":
    H_neighbors.append(edge[0])
    
sub_graph = G.subgraph(H_neighbors)


edge_weights = [sub_graph.edges[v,w]["weight"] for v,w in sub_graph.edges]
plt.figure(figsize=(20,10))
pos = nx.spring_layout(sub_graph,k=2)
nx.draw(sub_graph,pos=pos,node_size=20,node_color="#73000A",edge_color=edge_weights,with_labels=True)
node_0_position = pos["shafiur"]
plt.plot(node_0_position[0],node_0_position[1],'go',markersize=51 )

返回一个图，其中所有边和节点都连接到“shafiur”（sorted_eigenvector 中的第一个节点）。我想制作一个社交网络图，其中所有边和节点都连接出现在 sorted_eigenvector[:20] 中的所有节点，而不仅仅是 'shafiur'

解决方法

只需获取所有“特殊”节点的邻居，然后归纳出子图：

import networkx as nx

g = nx.karate_club_graph()

special_nodes = [0,5,22]

neighbors = []
for node in special_nodes:
    neighbors.extend(g.neighbors(node))

h = g.subgraph(special_nodes + neighbors)

pos = nx.spring_layout(g)

fig,axes = plt.subplots(1,2,sharex=True,sharey=True)
nx.draw(g,pos,with_labels=True,ax=axes[0])
nx.draw(h,ax=axes[1])
plt.show()