Jie Yang

University of Illinois at Chicago

D-optimal Designs with Ordered Categorical Data

We consider D-optimal designs with ordered categorical responses and cumulative link models. In addition to theoretically characterizing locally D-optimal designs, we develop efficient algorithms for obtaining both approximate designs and exact designs. For ordinal data and general link functions, we obtain a simplified structure of the Fisher information matrix, and express its determinant as a homogeneous polynomial. For a predetermined set of design points, we derive the necessary and sufficient conditions for an allocation to be locally D-optimal.

Thursday, March 26, 2015 - 3:30pm

Sandra Esi Safo

PhD Candidate, University of Georgia Department of Statistics

Design and Analysis Issues in High Dimension, Low Sample Size Problems

Advancement in technology and computing power have led to the generation of data with enormous amount of variables when compared to the number of observations. These types of data, also known as high dimension, low sample size, are plagued with different challenges that either require modifications of existing traditional methods or development of new statistical methods. One of these challenges is the development of Sparse methods that use only a fraction of the variables.

Major Professor(s): 
Dr. Jeongyoun Ahn and Dr. Kevin K. Dobbin
Monday, July 14, 2014 - 9:00am
The Cohen Room (230), Statistics Building

Emmanuel Tuglo

PhD Candidate, University of Georgia Department of Statistics

Analysis of Univariate and Multivariate Longitudinal Data with Censored and Missing Response with Complex Covariance Structure

The analysis of univariate and multivariate longitudinal data (U/MLD) with censored and missing response has inspired considerable interest in the statistical community recently. In the case of MLD, estimating the contemporaneous correlation coefficient is of particular interest to applied researchers.

Major Professor(s): 
Dr. Daniel Hall
Wednesday, July 2, 2014 - 10:00am
The Cohen Room (230), Statistics Building

Xiaoming Huo

National Science Foundation and Georgia Tech

Funding for Statistical Science at National Science Foundation, and a New Result on Dependence

This talk will have two components. In the first part, I will give an overview of NSF, and its resources that are relevant to the statistical science. Some related programs that are beyond the Division of Mathematical Science, such as the Big Data program and the Computational and Data-enabled Science and Engineering program, will be reviewed. Some suggestions regarding how to apply to these programs may be provided.

Tuesday, September 2, 2014 - 3:30pm

Ping-Shou Zhong

Michigan State University

Tests for High-dimensional Covariance Structures

Structured covariance matrices characterized by a small number of parameters have been widely used and play an important role in parameter estimation and statistical inference. To assess the adequacy of a specified covariance structure, one often adopts the classical likelihood-ratio test when the dimension of the data (p) is smaller than the sample size (n). However, this assessment becomes quite challenging when p is bigger than n, since the classical likelihood-ratio test is no longer applicable.

Thursday, December 4, 2014 - 3:30pm
Room 306, Statistics

Yichen Qin

University of Cincinnati

Robust Hypothesis Testing via Lq-Likelihood

In this talk, we introduce a robust testing procedure — the Lq-likelihood ratio test (LqLR).  We derive the asymptotic distribution of our test statistic and demonstrate its robustness properties both analytically and numerically.

Thursday, November 20, 2014 - 3:30pm

Weng Kee Wong


Using Animal Instincts to Find Efficient Experimental Designs

Experimental costs are rising and it is important to use minimal resources to make statistical inference with maximal precision. Optimal design theory and ideas are increasingly applied to address design issues in a growing number of disciplines, and they include biomedicine, biochemistry, education, agronomy, manufacturing industry, toxicology and food science, to name a few.  

Thursday, November 13, 2014 - 3:30pm

Jiming Jiang

UC Davis

On High-dimensional Misspecified Mixed Model Analysis in Genome-wide Association Study

We study behavior of the restricted maximum likelihood (REML) estimator under a misspecified linear mixed model (LMM) that has received much attention in recent gnome-wide association studies. The asymptotic analysis establishes consistency of the REML estimator of the variance of the errors in the LMM, and convergence in probability of the REML estimator of the variance of the random effects in the LMM to a certain limit, which is equal to the true variance of the random effects multiplied by the limiting proportion of the nonzero random effects present in the LMM.

Thursday, November 6, 2014 - 3:30pm

Lynne Billard

The University of Georgia

Distributions are the Numbers of the Future - Symbolic Data Analysis

Massively large data sets are routine and ubiquitous given modern computer capabilities. What is not so routine is how to analyse these data. One approach is to aggregate the data sets according to some scientific criteria. The resultant data are perforce symbolic data, i.e., lists, intervals, histograms, and so on. Applications abound, especially in the medical and social sciences. Other data sets (small or large in size) are naturally symbolic valued, such as species data, data with measurement uncertainties, confidential data, and the like.

Thursday, October 16, 2014 - 3:30pm

Christopher Nachtsheim

University of Minnesota

Recent Developments in Definitive Screening

Definitive Screening Designs (DSDs), discovered in 2011, are a new alternative to standard two-level screening designs. There are many desirable features of this family of designs. They require few runs while providing orthogonal main effects and avoiding any confounding of main effects by two-factor interactions. In addition they allow for estimating any quadratic effect of the continuous factors. The two-factor interactions are correlated but not confounded with each other. Moreover, in DSDs with 6 or more factors, it is possible to fit a full quadratic model in any three factors.

Thursday, October 23, 2014 - 3:30pm


Subscribe to RSS - Colloquium