Department Seminar: {Author}

Network analysis combines entity-based measurements (an individual's
behavior) with information about an entity's peers (a group's
behavior). This raises many questions; several analytical challenges
(e.g., rolling up group behavior to entity-level scores, breaking a
big network into small meaningful sub-networks etc.) as well as
practical challenges (e.g., connecting multiple entities seemingly
related). These challenges and approaches will be addressed through
two case studies: a fraud detection example in a social welfare
context, and contagious churn in a telecommunications marketing
context. In both contexts, the behavior of the entity's immediate
network can change that entity's predicted behavior.