Title:
Asymptotic Deficiency and Inadmissibility of Empirical Distribution Functions
Date:
Friday, November 18, 2011 from 2:00pm
Location:
E3-262 EITC Building, University of Manitoba Fort Garry Campus
Speaker:
Dr. Alexandre Leblanc
Department of Statistics
University of Manitoba
Abstract:
The estimation of distribution functions is one of the simplest problem of statistical function estimation. In this talk, I will discuss the fact that the empirical distribution function (also often referred to as the empirical process), perhaps one of the most frequently used tools in nonparametric statistics, is asymptotically inadmissible as an estimator of the underlying distribution function. In other words, it is possible to find other estimators that do better than it, for large sample sizes, according to reasonable measures of performance. Smooth estimators are especially interesting in this setting. I will also introduce the concept of asymptotic deficiency, which gives an interesting and somewhat different perspective on the problem.
Speaker Bio:
Alex Leblanc joined the Dept. of Statistics in July 2003. Before coming to Winnipeg, he obtained a Ph.D. in Statistics in the fall of 2002 from the Département de Mathématiques et de Statistique at the Université de Montréal. His thesis focused on the use of wavelets in Bayesian calculation and empirical Bayes estimation. He was promoted to Associate Professor and granted tenured in 2010.
His main research focus is now on statistical function estimation and nonparametric methods, including decision theoretical methods and asymptotics. Other recent interests are methods for sparse data and data depth.
Cost:
This will be a free event.
Contact:
For questions or more information: Jun Cai 474-6419
