Calculate average neighbor degree in networkX according to the attributes of the neighboring nodes

user_14083575 0
Viewed 1236

In my graph, the nodes have an attribute that tells me the type of node can be 1 or 2.

G=nx.Graph()
G.add_node('N1',n_type=1)
G.add_node('N2',n_type=1)
G.add_node('N3',n_type=2)
G.add_edges_from([('N3','N1'),('N3','N2')])

If I use the nx.average_neighbor_degree(G) function from NetworkX I get the following results.

{'N1': 2.0, 'N2': 2.0, 'N3': 1.0}

However, I would like to obtain the average neighbor degree considering only one type of node. For example, the average neighbor degree for the attribute when it takes the value of 1 should be:

{'N1': 0.0, 'N2': 0.0, 'N3': 1.0}

This, since node N3 is the only one that has neighbors with attribute n_type=1 and the average of it's neighbors is Degree(N1)+Degree(N2)/2 = 1+1/2 = 1

Any suggestions?

networkx python
1 Answers

Checking the source code and tweaking it a bit I could come up with this. Hope it could help.

import networkx as nx

def get_average_nbr_deg_by_type(G, node_type):
    avg = {}
    for n, deg in G.degree:
        nbrs_deg = [d for n, d in G.degree(G[n]) if G.nodes[n]['n_type'] == node_type]
        deg = len(nbrs_deg)
        if deg == 0:
            deg = 1
        avg[n] = sum(nbrs_deg) / float(deg)

    return avg

get_average_nbr_deg_by_type(G, node_type=1)
# outputs {'N1': 0.0, 'N2': 0.0, 'N3': 1.0}
user_124738630
Related