python生成连线图_python-有向图的最大强连接组件-CSDN博客

我正在根据总共82927个定向电子邮件数据构建的Networkx .MultiDiGraph()对象.在当前阶段,我试图从.MultiDiGraph()对象及其对应的子图中获取最大的强连接组件.

文本数据可以访问here.

这是我的工作代码：

import networkx as nx

import pandas as pd

import matplotlib.pyplot as plt

email_df = pd.read_csv('email_network.txt', delimiter = '->')

edge_groups = email_df.groupby(["#Sender", "Recipient"], as_index=False).count().rename(columns={"time":"weight"})

email = nx.from_pandas_dataframe(edge_groups, '#Sender', 'Recipient', edge_attr = 'weight')

G = nx.MultiDiGraph()

G.add_edges_from(email.edges(data=True))

# G is a .MultiDiGraph object

# using .strongly_connected_components() to get the part of G that has the most nodes

# using list comprehension

number_of_nodes = [len(n) for n in sorted(nx.strongly_connected_components(G))]

number_of_nodes

# 'number_of_nodes' return a list of [1, 1, 1,...,1] of length 167 (which is the exact number of nodes in the network)

# using the recommended method in networkx documentation

largest = max(nx.strongly_connected_components(G), key=len)

largest

# 'largest' returns {92}, not sure what this means...

正如我在上面的代码块中所指出的那样,列表理解方法返回一个长度为167 [[,1,1,…,1]的列表(这是我数据中节点的总数),而max( nx.strongly_connected_components(G),key = len)返回了{92},我不确定这是什么意思.

我的代码似乎有问题,我可能错过了处理数据的几个关键步骤.有人可以看一看并启发我吗？

谢谢.

注意：修改后的代码(对Eric和Joel表示敬意)

import networkx as nx

import pandas as pd

import matplotlib.pyplot as plt

email_df = pd.read_csv('email_network.txt', delimiter = '')

edge_groups = email_df.groupby(["#Sender", "Recipient"], as_index=False).count().rename(columns={"time":"weight"})

# per @Joel's comment, adding 'create_using = nx.DiGraph()'

email = nx.from_pandas_dataframe(edge_groups, '#Sender', 'Recipient', edge_attr = 'weight', create_using = nx.DiGraph())

# adding this 'directed' edge list to .MultiDiGraph() object

G = nx.MultiDiGraph()

G.add_edges_from(email.edges(data=True))

现在,我们检查该网络中最大的强连接组件(就节点数而言).

In [1]: largest = max(nx.strongly_connected_components(G), key=len)

In [2]: len(largest)

Out [2]: 126

最大的强连接组件包含126个节点.

[更新]

经过进一步的试验和错误,我发现将数据加载到networkx时需要使用create_using = .MultiDiGraph()(而不是.DiGraph()),否则,即使您为MultiDiGraph及其弱/紧密连接的子图,您仍然可能得到错误的边数！这将反映在您的.strongly_connected_subgraphs()输出中.

就我的情况而言,我将建议其他人使用这种单线

import networkx as nx

import pandas as pd

import matplotlib.pyplot as plt

G = nx.read_edgelist(path="email_network.txt", data=[('time', int)], create_using=nx.MultiDiGraph(), nodetype=str)

并且我们可以实现.strongly_connected_components(G)和strongly_connected_subgraphs进行验证.

如果您使用第一个代码块中的networkx输出G,则max(nx.strongly_connected_components(G),key = len)将给出具有126个节点和52xx边缘的输出,但是如果您应用上面列出的单线,你会得到：

In [1]: largest = max(nx.strongly_connected_components(G), key=len)

In [2]: G_sc = max(nx.strongly_connected_subgraphs(G), key=len)

In [3]: nx.number_of_nodes(G_sc)

Out [3]: 126

In [4]: nx.number_of_nodes(G_sc)