Heuristics in Analytics: A Practical Perspective of What Influences Our Analytical World

Chapter 7

Graph Analysis Case Studies

In the previous chapter we looked at an analytical new approach to performing behavior analysis. The most common approach to understanding individuals’ behavior, whether customers, subscribers, account or policy holders, tax payers, and so on, is to cluster the observations into groups based on similarity. It is quite common to do so in banking, telecom, retail, and insurance industries. Network analysis—or graph analysis—is an old discipline that allows us to deploy new analytical approaches, particularly when trying to understand group behavior. Based on graph analysis we can now look at the customers, subscribers, and account and policy holders not just in terms of who they are and which particular attributes describe them but also, and maybe most important, we can look at them in terms of how they behave when interacting with other individuals. It is a relationship approach to describing individual behavior.

Telecommunications has a definite, straight forward association between nodes and links and subscribers and calls (and texts). The way people use telecommunications services describes them better (for this particular purpose) than the individual attributes they may have, such as payment and usage history, products and services acquired, age, economic class, salary, gender, marital status, professional category, and so on. For mobile operators, it is more important to understand how their subscribers use and relate to each other than ...