-
Visualization
- Know your network
-
Five principles
- Show the data
- Maximize data-in ratio
- Erase non-data ink
- Erase redundant data-ink
- Revise and edit
-
In practice
-
What is interesting in the data
- Geography
- Communities
- Special significan nodes
-
Criteria
- Minimize edge crossing
- Uniform edge lengths
- Avoid nodes overlap
- The smaller and the sparser
-
Metrics
-
Degree
-
Edges
-
Directed
- indegree
- Non 0 entries
- On a given column
- Sum
- Outdegree
- Non 0 entries
- On a given row
- Sum
-
Undirected
- How many edges are incident on a node
-
Examples
- Who is the most popular?
- Coauthoring
-
Centrality
-
Examples
- Who your friends are
- Quality friends
- Well connected friends
- Inmediate network
-
Degree sequence
- Degree of each node
- Degree distribution
-
Other attributes
-
Weight
- Possitive
- Negative
- Ranking
- Type
- Graph structure
-
Is everything connected?
- Overall connectivity of the network
- Is everyone reacheble?
- Sparse
-
Connected components
-
Directed networks
- Strongly connected components
- Weakly connected components
-
Undirected networks
- Connected components
-
Giant component
- Significant portion of the graph
-
Data representation
-
Adjacency matrix
- Outgoing edges
-
Positions
- Rows
- Columns
- nr. of Node
-
Values
- Aij=1 ->i has an edge to j
- Aij=0 -> i has no edge to j
- Aij=Aji reciprocy
-
Aii=0 unless self-loops
- Diagonal=0
-
Edgelist
- Edges to
- Adjacency list