Department of Mathematical Sciences

University of Nevada, Las Vegas


Statistics Colloquium/Seminar Series

2010-2011

 

[2006-2008] [2008-2009] [2009-2010] [2011-2012] [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 2010


Friday, September 10
CBC-C224, 11:30 am

(refreshments at 11:15am)

Dr. Tapabrata Maiti
Department of Statistics

Michigan State University

Title: Computational and Inferential Issues in Disease Mapping Models
 

[Abstract] Mapping of small area mortality risks is a widely used technique in public health and in other area of statistical applications. The commonly used measure of risk, the standardized mortality ratio is not reliable due to its high variability in areas with low  population. Advanced statistical techniques, such as, hierarchical modeling is common to overcome this issue. However the spatially correlated structures often possess challenges for their implementation and inferences. In this talk we will discuss the statistical issues from frequentist and Bayesian perspective and will offer some new ways of solving the problem.



Friday, October 15
CBC-C224, 11:00 am

(refreshments at 10:45am)

Dr. Gary Aras
Clinical Development Biostatistics

Amgen Inc.

Title: A review of tests based on closed testing procedure
 

[Abstract] In clinical trials it is quite common to test more than one dose for efficacy against placebo or an active control arm. Regulatory agencies such as Food and Drug administration insist that if there are multiple comparisons to be made in a clinical trial then family-wise error rate should be controlled—typically at 5% level.   A closed testing procedure was developed by Marcus, Peritz and Gabriel and has been the mathematical foundation for multiple testing procedures. In general, we need to consider all possible intersections of the null hypotheses of interest.  A hypothesis is rejected if its associated test and all tests associated with hypotheses implying it are significant. We shall explore implications of this procedure to generate interesting tests that control family-wise error rate at a given level.



Friday, November 5
CBC-C224, 11:00 am

(refreshments at 10:45am)

Dr. Hokwon Cho 
Department of Mathematical Sciences

University of Nevada, Las Vegas

Title: Sequential Confidence Limits for the Ratio of Two Binomial Proportions
 

[Abstract] We present a sequential method for obtaining approximate confidence limits for the ratio of two independent binomial proportions based on a slightly modified maximum likelihood estimator. Large-sample properties of the proposed sequential estimator are studied. Monte Carlo simulation is carried out in order to investigate its finite sample behavior. The proposed method is applied to a numerical example for illustration and its use.

 

Spring 2011


Friday, February 4
CBC-C224, 11:00 am

(refreshments at 10:45am)

Dr. Dongseok Choi
Department of Public Health & Preventive Medicine

Oregon Health & Science University

Title: Detecting Subclusters in Outliers
 

[Abstract]  Medical research is often interested in finding subgroups in an outlier group.  For example, a certain medical condition can be more frequent in a small group that is different from the majority of population. One approach to find groups in a data set is using cluster analysis. Cluster analysis has been widely used tool in exploring potential group structure in complex data and has received greater attention in recent years due to data mining and high dimensional data such as microarrays. In this presentation, I will introduce split-and-recombine procedure and its application for a medical data set.  In addition, analysis results of the same data using other clustering methods will be discussed.



Friday, February 25
CBC-C224, 11:00 am

(refreshments at 10:45am)

Dr. S. Rao Jammalamadaka
Department of Statistics & Applied Probability

University of California, Santa Barbara

Title: Testing Isotropy and a related Random Walk problem
 

[Abstract]  One comes across directions as the observations in a number of situations. The first inferential question that one should answer when dealing with such data is, “Are they isotropic or uniformly distributed?” The answer to this question goes back in history which we shall retrace a bit and provide an exact and approximate solution to this so-called “Pearson’s Random Walk” problem.



Friday, March 11
CBC-C224, 11:00 am

(refreshments at 10:45am)

Dr. Jae-Kwang Kim
Department of Statistics

Iowa State University

Title: Parametric fractional imputation for missing data analysis
 

[Abstract]  In this talk, parametric fractional imputation is proposed as a frequentist approach of generating imputed values. Using the fractional weights, the E-step of the EM algorithm can be approximated by the weighted mean of the imputed data likelihood where the fractional weights are computed from the current value of the parameter estimates.  Some computational advantage over the existing methods can be achieved using the idea of importance sampling in the Monte Carlo approximation of the conditional expectation. The resulting estimator of the specified parameters can be identical to the MLE under missing data if the fractional weights are adjusted using a calibration step.



Friday, April 8
CBC-C224, 11:00 am

(refreshments at 10:45am)

Dr. Kaushik Ghosh
Department of Mathematical Sciences

University of Nevada, Las Vegas

Title: Modeling relational data using nested partition models
 

[Abstract]  This talk will introduce a flexible class of models for relational data based on a hierarchical extension of the two-parameter Poisson-Dirichlet process. The models are motivated by two different applications: 1) A study of cancer mortality rates in the U.S., where rates for different types of cancer are available for each state, and 2) the analysis of microarray data, where expression levels are available for a large number of genes in a sample of subjects. In both these settings, we are interested in improving the estimation by flexibly borrowing information across rows and columns while partitioning the data into homogeneous subpopulations. Our model allows for a novel nested partitioning structure in the data not provided by existing nonparametric methods, in which rows are clustered while simultaneously grouping together columns within each cluster of rows. The number of partitions are assumed to be unknown and are estimated from the data. We will illustrate our models using some real data examples.

 

 

è Statistics Colloquium/Seminar Series