Have you tried to plot a "real" graph, with many nodes? From my experience networkx does not assign the best layout to moderate and large graphs. For example the network of scientists http://networkdata.ics.uci.edu/data/netscience/netscience.gml with spring_layout or kamada_kawai_layout is either odd or with false connectivity.

I added draw_networkx_nodes_labels to draw_altair.py with the following added parameters:

• node_label_size : scalar or string Size of nodes (default=15). If an array is specified it must be the same length as nodelist.

• node_label_color : color string, or array of floats Node color. Can be a single color format string (default='r'), or a sequence of colors with the same length as nodelist. If numeric values are specified they will be mapped to colors using the cmap and vmin,vmax parameters. See matplotlib.scatter for more details.

• node_label : string The name of the node attribute to treat as a label.

Limited testing shows that it works as expected, though I haven't yet tested color.

Hi, I'm trying, without success, to assign a specific size to each node.

If I get it right this should be possible by linking the node_size property to a node attribute. Unfortunately, the result is a graph with all nodes represented as dots (size=1 i guess).

What am I doing wrong?

Here's my code:

for n in g.nodes():
    g.nodes[n]['name'] = n
    w = randint(300, 1000)
    g.nodes[n]['size'] = w
    print(w)
    
pos = nx.spring_layout(g)

viz = nxa.draw_networkx(
    g, 
    pos=pos,
    node_color='#ff0000',
    width='weight',
    edge_color='black',
    node_label='name',
    node_size='size'
)

viz.interactive()

added to_pandas_edges_arrows to core.py which calculates portions of the line to be shown as arrows.

added draw_networkx_arrows to draw_altair.py which draws the arrows as rectangles on the lines. Added parameters:

• arrow_width : float, optional (default=2.0) The width of arrow portions of edges.

• arrow_length : float, optional (default=.25) The proportion of the line to be occupied by the arrow.

Not dependent on my other pull request. Will require a merge to put them together, only conflict will be in draw_networkx due to the if statements.

Graphs look similar to a DiGraph in networkx base package plotting. Arrows are only appied if G isinstance nx.DiGraph.

This is a jupyter notebook that demonstrates how nx_altair can help interpret networks quickly. There is quite a bit of overlap with the existing example in terms of what nx_altair functionality is demonstrated, but I thought it might be helpful to have an example using a real dataset and using networkx functionality to determine colors and sizing.

Hi @Zsailer,

Thanks a lot for your work on this library.

It seems an Altair or Vega Light update has added default borders that might have not been included in prior versions. When I run the examples on Colab (on right side below) with current Altair 4.1, the network nodes receive distracting blue node outlines, which are not included in the ipynb files generated two years ago.

Is there any chance to make these node borders transparent or width=0?

All the best & thank you! Marian

Hi guys!

I've found a small issue with edge width. I've just fixed it, I think it is very clear from code what the problem was, but let me know if you want me to provide some examples before and after the fix.

As there is not develop branch or similar, I pointing to master, let me know if I should point other branch

Thansk!

HI,

The code below can be found in the method draw_networkx_nodes . I added a line that computes node_chart in the event a subset of nodes is specified via the nodelist argument. Without this line, the entire graph is drawn regardless of the nodelist argument.

   # ---------- Handle arguments ------------

    ###### node list argument
    if isinstance(nodelist, list):
        # Subset dataframe.
        df_nodes = df_nodes.loc[nodelist]
        # Build a chart (GE)  <<<<< ADDED THESE TWO LINES
        node_chart = alt.Chart(df_nodes)

In networkx >= 2.4, the Graph.node function has been deprecated in favour of Graph.nodes (see https://networkx.org/documentation/stable/release/release_2.4.html#deprecations).

This means that when using versions of networkx greater or equal to 2.4, when running the florentime_families example notebook, the code chunk pasted below results in an error:

for f in G.nodes():
    for i, c in enumerate(communities):
        if f in c:
            G.node[f].update({"community" : str(i),
                                "centrality" : centralities[f],
                                "name" : f
                               })

AttributeError: 'Graph' object has no attribute 'node'

It can be fixed by changing the bit that says G.node[f].update to G.nodes[f].update, though I'm not sure if a bigger change involving refactoring the code a little and using G.set_node_attributes() instead might be a better solution that works on all versions.

If it'd be helpful, I could submit a PR with one of the suggested fixes?

Thanks for the great work! I would like to use nxa.draw_networkx_edges to draw edges with a clorbar, like the graph at https://stackoverflow.com/questions/22124793/colorbar-for-edges-in-networkx. However, I got the following error:

SchemaValidationError: Invalid specification
altair.vegalite.v4.schema.core.Scale, validating 'additionalProperties'
Additional properties are not allowed ('legend' was unexpected)

I would greatly appreciate it if you could help. Thank you!

Best, Yongcheng

a = pd.DataFrame([{'a': 1, 'c': 0}, {'a': 2, 'c': 2}, {'a': 3, 'c': 5}]) b = a b = b.loc[[0, 2]]

b => a c 0 1 0 2 3 5

a => a c 0 1 0 1 2 2 2 3 5