APPLIED SOCIAL NETWORK

ANALYSIS IN PYTHON

Triadic Closure

Triadic closure: The tendency for

people who share connections in a social
network to become connected.

How can we measure the prevalence of

triadic closure in a network?
 APPLIED SOCIAL NETWORK
ANALYSIS IN PYTHON

Local Clustering Coefficient

Local clustering coefficient of a node:

Fraction of pairs of the node’s friends that are
friends with each other.
 APPLIED SOCIAL NETWORK
ANALYSIS IN PYTHON

Local Clustering Coefficient

Compute the local clustering coefficient of node C:
# of pairs of Cʹ s friends who are friends
# of pairs of Cʹ s friends

# of Cʹ s friends = 𝑑" = 4 (the “degree” of C)

# of pairs of Cʹ s friends =
 APPLIED SOCIAL NETWORK
ANALYSIS IN PYTHON

Local Clustering Coefficient

Compute the local clustering coefficient of node C:
# of pairs of Cʹ s friends who are friends
# of pairs of Cʹ s friends

# of Cʹ s friends = 𝑑" = 4 (the “degree” of C)

ʹ 𝑑" (𝑑" − 1)
# of pairs of C s friends =
2
 APPLIED SOCIAL NETWORK
ANALYSIS IN PYTHON

Local Clustering Coefficient

Compute the local clustering coefficient of node C:
# of pairs of Cʹ s friends who are friends
# of pairs of Cʹ s friends

# of Cʹ s friends = 𝑑" = 4 (the “degree” of C)

ʹ 𝑑" (𝑑" − 1) 12
# of pairs of C s friends = = = 6
2 2
 APPLIED SOCIAL NETWORK
ANALYSIS IN PYTHON

Local Clustering Coefficient

Compute the local clustering coefficient of node C:
# of pairs of Cʹ s friends who are friends
# of pairs of Cʹ s friends

# of Cʹ s friends = 𝑑" = 4 (the “degree” of C)

ʹ 𝑑" (𝑑" − 1) 12
# of pairs of C s friends = = = 6
2 2
# of pairs of Cʹ s friends who are friends =
 APPLIED SOCIAL NETWORK
ANALYSIS IN PYTHON

Local Clustering Coefficient

Compute the local clustering coefficient of node C:
# of pairs of Cʹ s friends who are friends
# of pairs of Cʹ s friends

# of Cʹ s friends = 𝑑" = 4 (the “degree” of C)

ʹ 𝑑" (𝑑" − 1) 12
# of pairs of C s friends = = = 6
2 2
# of pairs of Cʹ s friends who are friends = 2
 APPLIED SOCIAL NETWORK
ANALYSIS IN PYTHON

Local Clustering Coefficient

Compute the local clustering coefficient of node C:
# of pairs of Cʹ s friends who are friends
# of pairs of Cʹ s friends

# of Cʹ s friends = 𝑑" = 4 (the “degree” of C)

ʹ 𝑑" (𝑑" − 1) 12
# of pairs of C s friends = = = 6
2 2
# of pairs of Cʹ s friends who are friends = 2
2 1
Local clustering coefficent of C = =
6 3
 APPLIED SOCIAL NETWORK
ANALYSIS IN PYTHON

Local Clustering Coefficient

Compute the local clustering coefficient of node F:
# of pairs of F ʹ s friends who are friends
# of pairs of F ʹ s friends

𝑑. = 3
ʹ 𝑑. (𝑑. − 1) 6
# of pairs of F s friends = = = 3
2 2
# of pairs of Fʹ s friends who are friends = 1
1
Local clustering coefficent of F =
3
 APPLIED SOCIAL NETWORK
ANALYSIS IN PYTHON

Local Clustering Coefficient

Compute the local clustering coefficient of node J:
# of pairs of J ʹ s friends who are friends
# of pairs of J ʹ s friends

# of pairs of Jʹ s friends = 0 (Can not divide by 0)

We will assume that the local clustering coefficient

of a node of degree less than 2 to be 0.
 APPLIED SOCIAL NETWORK
ANALYSIS IN PYTHON

Local Clustering Coefficient

Local clustering coefficient in NetworkX:

G = nx.Graph()
G.add_edges_from([('A', 'K'), ('A', 'B'), ('A', 'C'), ('B', 'C'), ('B', 'K'),
('C', 'E'), ('C', 'F'), ('D', 'E'), ('E', 'F'), ('E', 'H'), ('F', 'G'), ('I', 'J')])

In: nx.clustering(G, 'F')

Out: 0.3333333333333333

In: nx.clustering(G, 'A')

Out: 0.6666666666666666

In: nx.clustering(G, ’J')

Out: 0.0
 APPLIED SOCIAL NETWORK
ANALYSIS IN PYTHON

Global Clustering Coefficient

Measuring clustering on the whole network:

Approach 1: Average local clustering coefficient

over all nodes in the graph.

In: nx.average_clustering(G)
Out: 0.28787878787878785
 APPLIED SOCIAL NETWORK
ANALYSIS IN PYTHON

Global Clustering Coefficient

Measuring clustering on the whole network (Approach 2):

Percentage of “open triads” that are triangles in a network.

Triangles: 3∗Number of closed triads

Transitivity =
Number of open triads

Open triads:
 APPLIED SOCIAL NETWORK
ANALYSIS IN PYTHON

Global Clustering Coefficient

Measuring clustering on the whole network:

Transitivity: Ratio of number of triangles and

number of “open triads” in a network.

In: nx.transitivity(G)
Out: 0.409090909091
 APPLIED SOCIAL NETWORK
ANALYSIS IN PYTHON

Transitivity vs. Average Clustering Coefficient

Both measure the tendency for edges to form triangles.
Transitivity weights nodes with large degree higher.

• Most nodes have • Most nodes have

high LCC low LCC
• The high degree • High degree node
node has low LCC have high LCC

Ave. clustering coeff. = 0.93 Ave. clustering coeff. = 0.25

Transitivity = 0.23 Transitivity = 0.86
 APPLIED SOCIAL NETWORK
ANALYSIS IN PYTHON

Summary
Clustering coefficient measures the degree to which nodes in a network
tend to “cluster” or form triangles.
Local Clustering Coefficient Global Clustering Coefficient
Fraction of pairs of the node’s Average Local Transitivity
friends that are friends with each Clustering Coefficient
other. Ratio of number of
triangles and number of
nx.average_clustering(G) “open triads”.
2 1
LCC of C = =
6 3
Puts larger weight on
high degree nodes.
nx.transitivity(G)