Is there a way to draw the graph "features" with networkx in Python?

Is there a way to draw the graph "features" with networkx ?

(-1, 0 and 1 in my example)

With my Python code, currently it does not draw it. See code below:

import torch
from torch_geometric.data import Data

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

#features: one feature for each node
#node features <------
x = torch.tensor([[-1], [0], [1]], dtype=torch.float)

data = Data(x=x, edge_index=edge_index)

do_visualize=1

if do_visualize:
    import networkx as nx
    import matplotlib.pyplot as plt

    edge_list = edge_index.t().tolist()
    
    G = nx.Graph()
    G.add_edges_from(edge_list)
    G.add_nodes_from(range(x.size(0)))

    pos = nx.spring_layout(G, seed=1)  # fixed seed for reproducibility
    
    fig=plt.figure(figsize=(6, 6))
    nx.draw_networkx(G, pos, with_labels=True, node_color='lightblue', edge_color='gray', node_size=800)
    plt.title("Visualization of the Social Network Graph")
    plt.axis('off')
    plt.show()

Nodes doesn't have information about your "features"

Primitive method: create dict with features = {nodename: text, ...} and use labels=features in draw_networkx() or use draw_networkx_labels()

features = {
        0: "[-1]",
        1: "[0]",
        2: "[1]",
}

# ...

nx.draw_networkx(G, pos, with_labels=True, ..., labels=features)

# or (with `with_labels=False`)

nx.draw_networkx(G, pos, with_labels=False)  # without labels=features
nx.draw_networkx_labels(G, pos, labels=features)

---

It may need better method to convert x to this dictionary.

I created only

features = dict(enumerate(x.int().tolist()))

or with extra text

features = {index: f"x = {item}" for index, item in enumerate(x.int().tolist())}

It needs to use .int() because it was creating strings like [1.0].

---

import torch
from torch_geometric.data import Data

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

#features: one feature for each node
#node features <------
x = torch.tensor([[-1], [0], [1]], dtype=torch.float)

data = Data(x=x, edge_index=edge_index)
#print(data.x)

features = {
        0: "[-1]",
        1: "[0]",
        2: "[1]",
}

#features = {index: str(item) for index, item in enumerate(x.int().tolist())}
#features = {index: item for index, item in enumerate(x.int().tolist())}
features = dict(enumerate(x.int().tolist()))

# or with extra text

#features = {index: f"x = {item}" for index, item in enumerate(x.int().tolist())}

print(features)

do_visualize = True

if do_visualize:
    import networkx as nx
    import matplotlib.pyplot as plt

    edge_list = edge_index.t().tolist()
    
    G = nx.Graph()
    G.add_edges_from(edge_list)
    G.add_nodes_from(range(x.size(0)))

    pos = nx.spring_layout(G, seed=1)  # fixed seed for reproducibility
    
    fig=plt.figure(figsize=(6, 6))

    nx.draw_networkx(G, pos, node_color='lightblue', edge_color='gray', node_size=800, with_labels=True, labels=features)

    # or `with_labels=False`

    #nx.draw_networkx(G, pos, node_color='lightblue', edge_color='gray', node_size=800, with_labels=False)
    #nx.draw_networkx_labels(G, pos, labels=features)

    plt.title("Visualization of the Social Network Graph")
    plt.axis('off')
    plt.show()

---

With converting to string you can format it in multilines and even use LaTeX

features = {index: f"x = {item}\ny = $2^{{ {item[0]} }}$" for index, item in enumerate(x.int().tolist())}

Doc: draw_networkx_labels, draw_networkx_edge_labels

See also how to convert data to graph

charts - How to visualize a torch_geometric graph in Python? - Stack Overflow

G = torch_geometric.utils.to_networkx(data, to_undirected=True)

but it also need to add labels separatelly.