Wenbo Wu

University of Georgia, Department of Statistics

SAS 2014 Data Analytics Shootout Presentation

The purpose of the 2014 SAS Data Analytics Shootout problem was to study the effect of various seasonal, economic, demographic, and climatological conditions on crime rates in five large US cities (Atlanta, Chicago, Denver, Houston, and Sacramento). Hourly precinct-level crime records for seven different crime types were provided for each city for different time periods between 2005 and 2012. Demographic and social economic data for each city were also provided, based on census data.

Thursday, March 5, 2015 - 3:30pm
Room 306, Statistics

Ying Xu

University of Georgia, Department of Biochemistry and Molecular Biology

Understanding Cancer in its Full Complexity through Mining Cancer Tissue Omic Data

A vast majority of the published cancer studies in the past few decades was conducted on cancer cells rather than cancer tissues. Knowing that the microenvironment plays key roles in cancer initiation, development and metastasis, we must reassess the true relevance of many of these published results to cancer. We have recently developed a new framework for cancer studies by treating cancer as a survival process under increasingly more challenging stresses, which evolve as a cancer evolves.

Thursday, February 26, 2015 - 3:30pm
Room 306, Statistics

Hubert Chen

University of Georgia

An Equivalence Test using the Fama-French Model and Its Application to US Stock Market
Hubert J Chen Professor Emeritus, UGA Department of Statistics Wen-Gine Wang Accountancy, NCKU, Taiwan

In this talk the three-factor Fama-French regression model (1992~1995) is introduced, where the three factors are the market risk premium (MRP), small-minus-big risk premium (SMB) and high-minus-low risk premium (HML).  It is known that the factors MRP, SMB and HML can affect a stock portfolio’s return.  According to the Fama-French regression model, their six types of stock portfolios created by company size (Small or Big) and its ratio of book-to-market equity (High, Medium or Low) can explain 90+% of the return for a portfolio.  

Thursday, February 19, 2015 - 3:30pm
Room 306, Statistics

Branislav Vidakovic

Georgia Institute of Technology - College of Engineering

Denoising by Bayesian Modeling in the Domain of Discrete Scale Mixing 2D Complex Wavelet Transforms

Abstract: Wavelet shrinkage methods that use complex-valued wavelets provide additional

insights to shrinkage process compared to standardly used real-valued

wavelets. Typically, a location-type statistical model with an additive noise is

posed on the observed wavelet coefficients and the true signal/image part is estimated

as the location parameter. Under such approach the wavelet shrinkage

becomes equivalent to a location estimation in the wavelet domain. The most

popular type of models imposed on the wavelet coefficients are Bayesian. This

Thursday, February 12, 2015 - 3:30pm
Room 306, Statistics

Harry Zhou

Yale University

Optimalities in Network Analysis

Network analysis is becoming one of the most active research areas in statistics. Significant advances have been made recently on developing theories, methodologies and algorithms for analyzing networks. However, there has been little fundamental study on optimal estimation. In this talk, we establish optimal rate of convergence for graphon estimation. For the stochastic block model with $k$ clusters, we show that the optimal rate under the mean squared error is $n^{-1}\log k+k^2/n^2$.

Thursday, March 19, 2015 - 3:30pm

Linwei Hu

PhD Candidate, University of Georgia Department of Statistics

Optimal Design of Experiments

In design of experiments, optimal designs are designs that  can glean the maximal amount of information from a study.   Therefore, an optimal design can reduce the number of ex- perimental units needed and saving the cost of study.   However, the research  in designing optimal experiments has not kept up with the increasingly complicated structure of data and models; especially for correlated data and multi-covariate models, finding optimal designs is very difficult.

Major Professor(s): 
John Stufken
Friday, November 14, 2014 - 10:30am
The Cohen Room (230), Statistics Building

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


Subscribe to RSS - Colloquium