使用Networkx 包对一个图分析与加工

使用Networkx 包对一个图分析与加工

第一步 导入图片

import networkx as nx
%matplotlib inline
import numpy as np
import matplotlib.pyplot as plt

simple_network = nx.Graph() #实例化一个空图
nodes = [1,2,3,4,5] #声明顶点
edges = [(1,2),(1,3),(1,4),(2,3),(2,5),(3,4)]#声明边
simple_network.add_nodes_from(nodes)# 加入顶点
simple_network.add_edges_from(edges)# 加入边
nx.draw(simple_network)# 画图

常用函数

输出顶点数

simple_network.order()
simple_network.number_of_nodes()
len(simple_network)

每个顶点相邻顶点们

simple_network.adjacency() 输出顶点
simple_network.degree() 输出相邻顶点数量

派系(任意两个顶点都相互连接)

from networkx.algorithms.clique import find_cliques
find_cliques(simple_network)

顶点与顶点间最短路径

nx.shortest_path(simple_network,6,8)
nx.shortest_path_length(simple_network,6,8)

第二步 添加顶点label

pos=nx.spring_layout(simple_network) # 返回一个字典，键为顶点，值为系统自动生成的顶点各自在图上的横纵坐标
# nodes
nx.draw_networkx_nodes(simple_network,pos,
                       node_color='r',
                       node_size=500,
                      alpha=0.8)
# edges
nx.draw_networkx_edges(simple_network,pos,
                       edgelist=edges,
                       width=8,alpha=0.5,edge_color='b')
# labels
node_name={}
for node in simple_network.nodes():
    node_name[node]=str(node)
nx.draw_networkx_labels(simple_network,pos,node_name,font_size=16)

plt.axis('off')# 去除图片边框
plt.show() # display

第三步 添加各个边label

赋予点和边内容与长度

address_list = [
    ('A',"Columbia University, New York, NY"),
    ('B',"Arco Cafe, Amsterdam Avenue, New York, NY"),
    ('C',"Riverside Church, New York, NY"),
    ('D',"Columbia Presbytarian Medical Center, New York, NY"),
    ('E',"Amity Hall Uptown, Amsterdam Avenue, New York, NY"),
]
distances = [
    ['A','B',10],
    ['A','C',5],
    ['A','D',25],
    ['A','E',7],
    ['B','E',3],
    ['D','E',21]
]

加入名为G_C的新图中

import networkx as nx
%matplotlib inline
G_C=nx.Graph()
node_labels=dict()
nodes = list()

for n in address_list:
    nodes.append(n[0])
    node_labels[n[0]] = n[1] 
for e in distances:
    G_C.add_edge(e[0],e[1],distance=e[2])#在加入edge时就添加一个distance的feature，存入距离

开始加label

pos=nx.spring_layout(G_C) # positions for all nodes

# nodes
nx.draw_networkx_nodes(G_C,pos,
                       node_color='r',
                       node_size=500,
                      alpha=0.8)
node_name={}
for node in G_C.nodes():
    node_name[node]=str(node)

nx.draw_networkx_labels(G_C,pos,node_name,font_size=16)
# edges
nx.draw_networkx_edges(G_C,pos,
                       edgelist=G_C.edges(),
                       width=8,alpha=0.5,edge_color='b')
nx.draw_networkx_edge_labels(G_C,pos,font_size=10)
plt.axis('off')
plt.show() # display

因为添加了距离，所以现在我们计算最短路径可以距离为权重

node1 = 'C'
node2 = 'E'
print(nx.shortest_path(G_C,node1,node2,weight='distance'))
print(nx.shortest_path_length(G_C,node1,node2,weight='distance'))