NETWORK ANALYSIS

What to explore in a Network?

Who is influential person in a particular
network

Network Flow

What are the parameters of judging the
influence

Influence is a single dimension parameter or
more than it?

Is there any locational advantage exist in N/W?

Network Structure useful for

business?

Whom to target for online
marketing campaign ?

Who can increase my reach?

Who is more influential ?

How to reach him ?

How to make promotional event
successful?
3

Locational Advantage /
Disadvantage

Locational Advantage /
Disadvantage
A

G
E

Locational Advantage / Disadvantage

E

F
D

Networks and Power:

1
2

3
1

3
4
4
5
5

Who has more Power?

Network Representation

Actors (Nodes, vertices or points)
 Represented as points
 Position of actor : Favourable position
v/s constraint position

Relationship (Edges, Arcs, Lines, Ties)

Dyad : A tie between two actors

Triad: A relationship between three
actors
8

Continued..

Graph
 Directed / undirected

Subgroup : A small group of few

members of any network

Centrality

Diameter

Modularity
9

Network Structure and Data Sources

Actors- Nodes or Vertex

NETWORK REPRESENTATION
B

NETWORK REPRESENTATION
B

Network Data
In general, a relation can be:
Binary or Valued
Directed or Undirected
b

d
4

Undirected, Valued

Directed, binary

Undirected, binary

c
Directed, Valued

Network Structure and Data Sources

Relations- Ties, Edges, arcs, Links

Undirected

Network Structure and Data Sources

Relations- Ties, Edges, arcs, Links
Undirected
Weight

Network Elements
Relations- Ties, Edges, arcs, Links
Directed

Basic Data Structures

From matrices to lists
a b c d e
a
1
b 1
1
c
1 1
1
d
1
1
e
1 1

Adjacency List
ab
bac
cbde
dce
ecd

Arc List
ab
ba
bc
cb
cd
ce
dc
de
ec
ed

Network Structure

How To Define the Power of a

particular Node in a Network ?

Concepts of Centrality

Diameter : Longest distance between
any pair in the network

Density : How the individual nodes
sharing information
 High density more connections quick
sharing of information

Network Centrality

Degree Centrality
Closeness Centrality
Betweenness Centrality
Eigen vector Centrality

Degree Centrality

Number of Links attached to a
particular node

More links more connections

Directed /Undirected links

In degree Number of links directed
to the node

Out degree Number of links
directed from that node

Degree Centrality ..

High in-degree More people are
following that node (more popular)

High out degree gregariousness
and social / generous

22

Degree Distribution
A

A =2
B =2
C =4
D =1
E =1

Degree Distribution

2.5

2
1.5
1
0.5
0
1

Degree

Degree Distribution
3

4
4

Degree Distribution
3

2
Degree Distribution
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0

4
4

DEGREE

Degree Centrality
J

I
H

R=

F=

D=

B=

A
Q

N
M

P
D
O

B
C

POWER LAW DISTRIBUTION

Which Node is Most

Important?
B

A
J

R
L
N

M
P
O

F
E

Which Node is Most Important?

J

I
H

A
Q

N
M

P
D
O

B
C

Closeness Centrality

It defines the proximity of the nodes

Distance between two nodes is
expressed as the number of links
present between them
 Actors who can reach to other actors in
shorter path lengths are at favoured
positions

Shortest path How easily one can

reach to other

Smaller number is better

Closeness Centrality
I
J

G
R

P
F

=18/
10

1.8

Closeness Centrality
I
J

G
R

P
F

=19/
10

1.9

Between ness Centrality

Between-ness Centrality

Measure of Nodes Influence

The number of shortest paths from all
vertices to all others that pass
through that node.

Degree to which a particular node is
work as a gate keeper to the network

High betweenness more important
the node is.

Betweenness Centrality

The degree to which a actor lies
between other nodes

The data will pass through in between
node

Actors with high betweenness can serve
as broker agent and play important role
in transfer of data

In- between node failure will cease the
communication between the nodes

This measure can be calculated in
absolute value, as well as in percentage

Between ness Centrality

A

B
E

C
D

F
H

Eigen Vector Centrality

Measure of nodes importance.

Iterative matrix computation that
gives more weight to nodes if they
are connected to influential nodes.

The backbone to techniques like
Googles PageRank which ranks web
pages.

Weight to the links

DENSITY
EDGES: 5

A
A
D
B

E
C

DENSITY

Nodes: 8
Edges: 12
Total Possible
Edges: ??

DENSITY

Nodes: 8
Edges: 12
Total Possible
Edges: ??
# Nodes * (# Nodes 1)
2

DENSITY

Nodes: 8
Edges: 12
Total Possible
Edges: ??

# Nodes * (# Nodes 1 )
2
(8 * 7 ) 56
=
= 28
2
2

Density:
12/28=0.43

Network Modularity and

Community Identification


Modularity : Subgroups in a big group

Subgroups are having high density
among itself.

Giant component : Largest
component of all the connected nodes

Giant component : used to filter out
the network

43

NETWORK MODULARITY
Merry
Mike
Mia
Bob

Emma

LIZ

Allen

Lisa

Jill
Mark
Shane

John

Tom

Leah
Jen

Sean

NETWORK MODULARITY
Merry
Mike
Mia
Bob

Emma
LIZ

Allen

Lisa

Jill
Mark
Shane

John

Tom

Leah
Jen

Sean

NETWORK MODULARITY
Merry
Mike
Mia
Bob

Emma

LIZ

Allen

Lisa

Jill
Mark
Sha
ne

John

Tom

Leah
Jen

Sean

GAINT MODULARITY
Merry
Mike
Bob

Mia
Emma

LIZ

Allen

Lisa

Jill
Mark
Shane

John

Tom

Leah
Jen

Sean

NETWORK MODULARITY
Merry
Mike
Bob

Mia
Emma

LIZ

Allen

Lisa

Jill
Mark
Shane

John

Tom

Leah
Jen

Sean

Connectivity and
Cohesion
Minimum number of nodes to remove before the
network become disconnected
A

F
B

A
C

F
G

E
Cohesion
=1

G
Cohesion
=2

50