Wenxuan Zhong

COP: a forward selection procedure for the index model

In this talk, I will present the Correlation Pursuit (COP) method, a variable selection procedure developed under the sufficient dimension reduction framework. Unlike the conventional stepwise, COP does not impose a special form of relationship between the response variable and the predictor variables. The COP procedure selects variables that maximize the correlation between the transformed response and linear combinations of the predictors.

Thursday, February 23, 2012 - 3:30pm
Type: 

Xuming He

Bivariate Downscaling for Climate Projections

Statistical downscaling is a useful technique to localize global or regional climate model projections to assess the potential impact of climate changes. It requires quantifying a relationship between climate model output and local observations from the past, but the two sets of measurements are not necessarily taken simultaneously, so the usual regression techniques are not applicable. In the case of univariate downscaling, a simple quantile-matching approach with asynchronous measurements often works well, but challenges remain for downscaling bivariate data.

Thursday, February 16, 2012 - 3:30pm
Type: 

M. Reza Meshkani

Shahid Beheshti University

Analysis of Covariance under Inverse Gaussian Model

In this talk, the limitations of normal model for Analysis of Covariance for positive right-skewed variables are considered. Specifically, an Inverse Gaussian variable is considered whose variance depends on its mean thus violating the usual assumptions of Normal linear model. Instead of appealing to transformations which makes interpretations of the results awkward, we propose a method of direct statistical analysis from both Maximum Likelihood and Bayesian perspectives. The formulas for adjusting treatment effects are given and their properties are discussed.

Tuesday, April 17, 2012 - 3:30pm
Type: 

Antony Joseph

Yale University

Near-optimal recovery using iterative algorithms in high-dimensional regression with random designs

We provide theoretical analysis of iterative algorithms for two problems in high-dimensional regression. In the first, a sparse linear model with a specific coefficient structure provides a framework for a problem in communication. We show that the algorithm has optimal performance when compared to information-theoretic limits. This provides theoretically provable, low computational complexity communication systems based on our statistical framework.

Thursday, February 2, 2012 - 3:30pm
Type: 

Pengsheng Ji

Cornell University

UPS Delivers Optimal Phase Diagram in High Dimensional Variable Selection

Consider a linear model Y = X + z, z N(0; In). Here, X = Xn;p, where both p and n are large, but p > n. We model the rows of X as i.i.d. samples from N(0; 1 n ), where is a pp correlation matrix, which is unknown to us but is presumably sparse. The vector is also unknown but has relatively few nonzero coordinates, and we are interested in identifying these nonzeros. We propose the Univariate Penalization Screeing (UPS) for variable selection. This is a screen and clean method where we screen with univariate thresholding, and clean with penalized MLE.

Thursday, January 26, 2012 - 3:30pm
Type: 

Kehui Chen

University of California, Davis

Modeling Conditional Distributions for Functional Data and Stringing High-Dimensional Data to Functions

In this talk, I will first present a method for conditional distribution and quantile estimation when predictors take values in a functional space, which is an extension of the usual functional mean regression. The study is motivated and illustrated by an application to the assessment of children’s growth patterns. The proposed method is supported by theory and is shown to perform well in simulations. An extension of the proposed conditional approach to model the more complex case when responses are also functions will be briefly discussed.

Tuesday, January 24, 2012 - 3:30pm
Type: 

Roger Woodard

A Large Course Redesign of Introductory Statistics: You don't need 65 other people with you to learn a definition.

Statistics 311 at NCSU is a large introductory course that serves over 700 students per semester. In this talk I will present the results of a redesign that transformed this course in a hybrid format. This project converted two of the three hours of student contact to an online format. Students still come together one day a week to participate in discussions where an instructor teaches them the important ideas and concepts. During this time students also participate in group activities that illustrate key ideas.

Thursday, January 19, 2012 - 3:30pm
Type: 

Luyan Dai

Boehringer Ingelheim Pharmaceuticals, Inc.

Empirical Bayesian Methods for Enrollment and Event Projection in Oncology Trials

In oncology trials with time to event as primary endpoints, the trial duration is mainly driven by the number of events observed to ensure sufficient power to draw confirmatory conclusions. The trial planning determined by clinical staff and the deterministic techniques fails to account for the uncertainties and stochastic fluctuations in the recruitment process and event evolvement. In this project, we incorporate a Poisson-gamma recruitment process into an exponential event prediction model under the empirical Bayesian setting.

Thursday, January 12, 2012 - 3:30pm
Type: 

Warren Kuhfeld

SAS Institute

Recent Directions in SAS Statistical and Graphical Software

This presentation begins with a behind-the-scenes look at how research statistician developers at SAS interact with customers and the statistical community to decide on new functionality in SAS/STAT. The presentation then provides a high-level tour of new directions and features in the most recent releases of SAS/STAT (9.2, 9.22, and 9.3).

Friday, October 21, 2011 - 9:00am
Type: 

Pages

Subscribe to RSS - Colloquium