Networkx：从Pandas DataFrame中的距离矩阵绘制网络

Question

I have a cosine distance matrix as a pandas dataframe. I can visualise it as a network if I convert it to a numpy matrix and do the following:

DistMatrix = cosine1
G = G=nx.from_numpy_matrix(DistMatrix)
nx.draw(G , with_labels=True)
plt.show()

The problem is that I loose the labels of my nodes. These labels were in the pandas dataframe however. Being a distance matrix, the column labels are of course the same as the index labels and the diagonal of the matrix has entries which are 0 (they are exactly the same).

How would I draw a network from my pandas dataframe and include the labels to label the nodes correctly?

 DistMatrix = pd.DataFrame( data = cosine , index = label_list , columns = label_list)

Is it maybe possible to just add labels to the numpy matrix version?

Answer 1

You can add labels manually with the function relabel_nodes by adding a dictionary with the mapping of the numbering to the labels.

G = nx.from_numpy_matrix(DistMatrix.values)

labels = DistMatrix.columns.values
G = nx.relabel_nodes(G, dict(zip(range(len(labels)), labels)))

nx.draw(G, with_labels=True)
plt.show()