What is the statistical modeling of extremes?

Speaker: Troy Wixson

Abstract: Much of statistics focuses on the bulk of the distribution where things like averages and correlation are sensible. In many systems the events that dictate the character of the system are the extreme events. These extreme events can be weather disasters like wildfires or flooding, massive file sizes in internet traffic, or extreme losses across financial portfolios. In this talk I will briefly introduce the statistical frameworks commonly used to model the frequency and intensity of extreme events like these. We will move through the well-developed univariate frameworks into the multivariate frameworks. While discussing the multivariate frameworks we will touch on several open challenges in the field and some of the work I have been doing in that space including time series modeling of wildfires, neural-network based classification, and the development of novel estimation methods.