Department of Mathematical Sciences

University of Nevada, Las Vegas


Statistics Colloquium/Seminar Series

2009-2010

 

[2006-2008] [2008-2009] [2010-2011] [Current Year]

 

For more information, contact the Colloquium/Seminar Coordinator, Dr. Hokwon Cho

(To see Math Dept Colloquia/Seminars, click next: Math Dept Seminar)

 

Fall 2009

 

·         Fri. 1:00 p.m. September 25, SEB-1240:  Dr. Peter Müller

Department of Biostatistics,  University of Texas, M.D. Anderson Cancer Center, Houston
Title: Modeling Dependent Gene Expression

Abstract: We consider statistical inference for high throughput genomic data. Most traditional statistical methods implicitly assume independent sampling (conditional on some hyperparameters). Recognizing the limitations of independent modeling we develop a model that includes a simple dependence structure across genes (or proteins). The important features of the proposed model are the ease of representing typical prior information on the nature of dependencies, model-based parsimonious representation of the signal as a ordinal outcome, and the use of a coherent probability model over both, structure and strength of the conjectured dependencies.
As part of the inference we reduce the recorded data to a trinary response representing underexpression, average expression and overexpression. For proteins the trinary response is further reduced to a binary indicator for activation. To achieve this, we use an extension of a model proposed in recent literature. Inference in the described model is implemented through a straightforward Markov chain Monte Carlo (MCMC) simulation, including posterior simulation over conditional dependence and independence. We use the proposed dependence probability model to derive inference about molecular pathways, including differential pathway activation across biologic conditions.

 

·         Fri. 12:30 p.m. November 4, SEB-1240:  Dr. Ching-Ti Liu

Department of Biostatistics,  University of Boston
Title: A Forest-Based Approach in Genome-Wide Association Study

Abstract: Multiple genes, gene-by-gene interactions, and gene-by-environment interactions are believed to underline most complex diseases. However, such interactions are difficult to identify.  While there have been recent successes in identifying genetic variants for complex diseases, it still remains difficult to identify gene-gene and gene-environment interactions.  To overcome this difficulty, we propose a forest-based approach and a concept of variable importance.  The proposed approach is demonstrated by simulation study for its validity and illustrated by a real data analysis for its use. Analyses of both real data and simulated data based on published genetic models show the effectiveness of our approach.

 

·         Fri. 12:30 p.m. November 20, SEB-1240:  Dr. Donatello Telesca

Department of Biostatistics, University of California, Los Angeles
Title: Functional Mixed Registration Models

Abstract: Functional data often exhibit a common shape but also variations in amplitude and phase across curves. The analysis often proceeds by synchronization of the data through curve registration. We propose a Bayesian hierarchical model for curve registration. Our methodology is extended to define a class of probability models, which combine curve registration with functional mixed effects modeling, discriminating phase and amplitude variability in a joint fashion. We discuss this class of models with a focus on penalized smoothing splines and propose Bayesian inferential procedures based on Markov Chain Monte Carlo samples from the posterior distribution of the functions of interest. We illustrate the application of our model using simulated data as well as to two datasets, namely, the Berkeley study on human growth and a study on the pharmacokinetics of the drug Remifentanil. Time permitting, we will introduce a generalized view of curve registration with applications to longitudinal counts of criminal activity.

 

Spring 2010

 

·         Fri. 1:30 p.m. February 19, SEB-1245:  Dr. Jeffrey Rosenthal

Department of Statistics,  University of Toronto
Title: Optimizing and Adapting the Metropolis Algorithm

Abstract: The Metropolis algorithm is a very popular method of approximately sampling from complicated probability distributions, such as those arising in Bayesian inference.  A wide variety of proposal distributions and tunings are available, and it can be difficult to choose among them. One possibility is to have the computer automatically "adapt" the algorithm while it runs, to improve and tune on the fly.  However, natural-seeming adaptive schemes can destroy the ergodicity properties which are essential for the algorithm to be valid.  In this talk, we first discuss general principles about optimizing Metropolis algorithms. We then consider adapting the algorithm, and explain using a very simple graphical example (probability.ca/jeff/java/adapt.html) how ergodicity can fail.  We present a theorem which gives simple conditions that ensure ergodicity, and consider several high-dimensional adaptive Metropolis and Metropolis-within-Gibbs examples.  Finally, we briefly discuss a preliminary general-purpose adaptive MCMC software package (probability.ca/amcmc).  Much of this is joint work with G.O. Roberts.

 

·         Fri. 1:30 p.m. February 26, SEB-1245:  Dr. Andrew Martin

School of Law and Department of Political Science, Washington University in St. Louis

Title: Untangling the Causal Effects of Sex on Judging

Abstract: We explore the role of sex in judging by addressing two questions of long-standing interest to political scientists:  whether and in what ways male and female judges decide cases distinctly---``individual effects"---and whether and in what ways serving with a female judge causes males to behave differently---``panel effects." While we attend to the dominant theoretical accounts of why we might expect to observe either or both effects, we do not use the predominant statistical tools to assess them. Instead, we deploy a more appropriate methodology: semi-parametric matching, which follows from a formal framework for causal inference.  Applying matching methods to thirteen areas of law, we observe consistent gender effects in only one---sex discrimination. For these disputes, the probability of a judge deciding in favor of the party alleging discrimination decreases by about 10 percentage points when the judge is a male. Likewise, when a woman serves on a panel with men, the men are significantly more likely to rule in favor of the rights litigant. These results are consistent with an informational account of gendered judging and are inconsistent with several others.  Time permitting, I will discuss the Supreme Court Database websites.

 

·         Fri. 1:30 p.m. April 16, SEB-1245:  Dr. Ian McKeague

Department of Biostatistics, Columbia University.

Title: Fractals and finding the locations of differentially expressed genes

Abstract: The statistical literature on gene expression data has mostly been concerned with multiple testing procedures for selecting sets of candidate genes that are differentially expressed between cases and controls.  This talk discusses the complementary problem of finding confidence regions for the locations (along a chromosome) of such genes.  Bootstrap methods for addressing this problem are introduced. The asymptotic performance of the proposed approach is studied in a functional data analysis framework in which the gene expression profiles are treated as stochastic processes having certain fractal properties.

 

·         Fri. 12:00 p.m. April 30, SEB-1245:  Dr. Adrian Dobra

Department of Statistics,  University of Washington, Seattle
Title: Dynamic Markov Bases

Abstract: We present a computational approach for generating Markov bases for multi-way contingency tables whose cells counts might be constrained by lower and upper bounds. Instead of computing the entire Markov basis in an initial step, this framework finds sets of local moves that connect each table in the reference set with a set of neighbor tables. We construct a Markov chain on the reference set of tables that requires only a set of local moves at each iteration. The union of these sets of local moves forms a dynamic Markov basis. We illustrate the practicality of these algorithms in two numerical examples.

 

·         Thu. 2:00 p.m. May 13, CBC C-122:  Dr. Sam Weerahandi

 Pfizer Inc.
Title: Response Estimation via Mixed Models With Application to Pharmaceutical Promotions

Abstract:  In estimating response by a large number of Segments or Subjects, simple regressions yield highly unreliable estimates.  The problem can be alleviated by using mixed models. Then BLUP of segment estimates become a function of variance components, which are traditionally estimated by MLE. It is well known that MLE could yield negative and/or unreliable estimates for variance components.  We proposed a generalized inference based estimation method which is substantially superior to MLE in this class of applications.  The utility of the method will be demonstrated involving a pharmaceutical marketing application where one needs to estimate hundreds of thousands of parameters. A demo of a tool that utilizes estimated responses in Profit Maximization will be given with some hypothetical and yet representative data.

 


 

è Statistics Colloquium/Seminar Series