April 18, 2014

Fields-Carleton Distinguished Lecture Series
Jerry Lawless,
Department of Statistics and Actuarial Science
University of Waterloo

Statistics and the Science of Risk
Thursday March 25, 2010
4:30-5:30 p.m.

Modeling and Analyzing Event Histories
Friday March 26, 2010
10:30-11:30 a.m.

Carleton University, School of Mathematics and Statistics
4351 HP (Macphail Room)

There will be a reception after the general public talk on Thursday at 5:30, and coffee will be served on Friday at 10:00 a.m.(before the specialized talk).

Talk 1 (General)

Statistics and the Science of Risk

As a society and as individuals we continually assess and discuss risk. Psychological aspects of risk perception and tolerance, and individual or corporate utilities associated with specific events and outcomes, play important roles in decisions concerning risks. However, at the core of scientific risk analysis are probabilities assigned to future eventualities. Statistics is fundamental to this aspect of risk, in part because of its connections with probability but more importantly in most settings, because it is the framework for learning about complex processes and environments so that probabilities can sensibly be assigned. This talk will discuss the role of statistics, beginning with an historical review of probability, statistics and risk and then moving to some current issues. Illustrations will be drawn from areas ranging from finance and gambling to health.

Talk 2 (More specialized)

Modeling and Analyzing Event Histories

Many settings involve events that occur over time for individuals or systems; for example, individuals can experience episodes of illness or hospitalization, software systems may experience failures, and insurers can receive claims from their customers. Event processes are typically modeled and analyzed in two main ways: dynamically, by considering the probability of a new event at a given time, conditional on the prior history of events; and what I will term pattern-wise, by considering features such as the expected numbers and distributions of events in various time intervals. In this talk I will discuss pros and cons of the two approaches and survey methods of analysis. The emphasis will be on models and applications of methodology, with examples. Technical details will be mentioned but downplayed.

back to top