Node-level metrics

Analyzing Social Media Data in Python

Alex Hanna

Computational Social Scientist

Centrality: node importance

Centrality
- Measures of importance of a node in a network
- Several different ideas of "importance"

Degree centrality

Degree

Number of edges that are connected to node
Two types of degrees in a directed network
- In-degree - edge going into node
- Out-degree - edge going out of a node

nx.in_degree_centrality(T)
nx.out_degree_centrality(T)

In-degree centrality

Betweenness centrality

How many shortest paths between two nodes pass through this node
Importance as a network broker

nx.betweenness_centrality(T)

Betweenness centrality

Printing highest centrality

bc = nx.betweenness_centrality(T)

betweenness = pd.DataFrame(
      list(bc.items()), 
      columns = ['Name', 'Cent'])

print(betweenness.sort_values(
       'Cent', 
       ascending = False).head())

    Name  Centrality
0      0    0.232540
23    23    0.158514
7      7    0.158514
15    15    0.158514
21    21    0.157588

Betweenness centrality

Centrality in different networks

Centrality 2x3 table

The ratio

degree_rt = pd.DataFrame(list(G_rt.in_degree()), 
                         columns = ['screen_name', 'degree'])
degree_reply = pd.DataFrame(list(G_reply.in_degree()),
                            columns = ['screen_name', 'degree'])

ratio = degree_rt.merge(degree_reply, 
                        on = 'screen_name', 
                        suffixes = ('_rt', '_reply'))

ratio['ratio'] = ratio['degree_reply'] / ratio['degree_rt']

Let's practice!

Analyzing Social Media Data in Python