CMPS 163: Business Analytics
Introduction
After looking at looking a k-means clustering in an earlier module, this time we will look at clustering again but we will use a different approach. Network graphs consist of nodes that are connected by edges, for example, people on Facebook could be nodes and if they are friends they have an edge between them. Community detection looks for communities within network graphs, and one approach to this is modularity maximization. Modularity maximization puts nodes that are well connected to each other in the same cluster, and all the other nodes in other clusters. We will see how this is calculated in detail. This is a vastly different approach from k-means, because it uses network graphs and does not explicitly represent the cluster centers. Besides implementing modularity maximization in Excel, we will also look at Gephi which is software that has nice visualizations of clusters and also implements modularity maximization.
Module Objectives
- Define network graphs, community detection, and modularity maximization
- Interpret an adjacency matrix
- Import data into Gephi
- Locate basic functionality in Gephi
- Prepare an r-neighborhood adjacency graph
- Implement community detection in Excel
- Run community detection in Gephi
Learning Resources
- Module 10 Readings: First half of Chapter 5
- Module 10 Slides: First half of Chapter 5
Learning Activities
- Module 10 Assignment
Videos
The friends example:
Clustering part one:
For Further Study
- Read more about cluster analysis on Wikipedia
- Read more about community detection on Wikipedia
- Read more about modularity on Wikipedia
- If you are into Game of Thrones, check out Network of Thrones!