Pytorch-Geometric

vision
1

I’m new to Pytorch-geometric, and geometric deep learning. I’m trying to visualize the datasets available in pytorch-geometric, but couldn’t find anything to do so.

How do I visualize these graph datasets. Is there provision to do so in pytorch-geometric? How else can we visualize this ?

1 Like
2 
from torch_geometric.datasets import KarateClub
dataset = KarateClub()

for i in dataset[0]:
  print(i)
# this torch.geometric.datasets object comprises of edge(edge information for each node), x(nodes) and y(labels for each node)

edge,x,y = dataset[0]
numpyx = x[1].numpy()
numpyy = y[1].numpy()
numpyedge = edge[1].numpy()

import networkx as nx

g = nx.Graph(numpyx)

name,edgeinfo = edge

src = edgeinfo[0].numpy()
dst = edgeinfo[1].numpy()
edgelist = zip(src,dst)

for i,j in edgelist:
  g.add_edge(i,j) 

nx.draw_networkx(g)
3

Hi,
What is nx here? Is it networkx library?

4

Yes. My apologies. I’ve now modified the code to include the import line.

5

No problem, Good luck.

6

Actually… there’s an even better way. PyG has something in-built to convert the graph datasets to a networkx graph.

import networkx as nx
import torch
import numpy as np
import pandas as pd
from torch_geometric.datasets import Planetoid
from torch_geometric.utils.convert import to_networkx

dataset1 = Planetoid(root = '/content/cora',name='Cora')

cora = dataset1 [0]

coragraph = to_networkx(cora)

node_labels = cora.y[list(coragraph.nodes)].numpy()

import matplotlib.pyplot as plt
plt.figure(1,figsize=(14,12)) 
nx.draw(coragraph, cmap=plt.get_cmap('Set1'),node_color = node_labels,node_size=75,linewidths=6)
plt.show()
4 Likes
7

Hello all! I am also new in PyG… I am trying to import the dataset using Colab but following your instructions outputs an error: module ‘community’ has no attribute ‘best_partition’. Do you know why this could be happening? Thanks for the help!!

from torch_geometric.datasets import KarateClub

dataset = KarateClub()
AttributeError: module ‘community’ has no attribute ‘best_partition’

8

For future issues, python-louvain already installs networkx package. Therefore, if the previous error appears uninstall both libraries and install back only python-louvain.

9

this is impler:

import networkx as nx

edge_index = torch.tensor([[0, 1, 1, 2],
                           [1, 0, 2, 1]], dtype=torch.long)
x = torch.tensor([[-1], [0], [1]], dtype=torch.float)

data = torch_geometric.data.Data(x=x, edge_index=edge_index)
g = torch_geometric.utils.to_networkx(data, to_undirected=True)
nx.draw(g)

you need matplot lib installed too do conda install -y matplotlib.