Spring 2010

Spring courses

Graphical schedule of courses-pdf

Fall courses

Fall 2010

Course descriptions

• ST 505 001 Applied Nonparametric Statistics (Pam Arroway)
  The course provides an introduction to statistical methods requiring relatively mild assumptions about the form of the population distribution. Hypothesis testing, point and interval estimation and multiple comparison procedures for a variety of statistical problems are covered.
• ST 610 001 Applied Data Mining (Dave Dickey); S/U grading
  Data mining is a collection of statistical tools appropriate for very large data sets that are typically from observational studies. Statistical significance plays a minor role here and other evaluation techniques like lift charts and ROC curves are the usual evaluation tools. The main technique, recursive splitting, is often used in market segmentation. The idea here is to use measurements on individuals, so called features, to divide a population for which these features are available into subsets that differ by, in a banking example, their probabilities of defaulting on debt. Logistic regression , regression trees and classification trees are the main techniques. Neural networks, clustering, market basket analysis and discriminant analysis round out the collection of tools covered and throughout the use of industrial strength software, currently SAS Enterprise Miner, is emphasized.
• ST 731 001 Applied Multivariate Analysis (Lexin Li)
  An introduction to use of multivariate statistical methods in analysis of data collected in experiments and surveys. Topics covered including multivariate analysis of variance, discriminant analysis, canonical correlation analysis and principal components analysis. Emphasis upon use of a computer to perform multivariate statistical analysis calculations.
• ST 732 001 Applied Longitudinal Analysis (Wenbin Lu)
  Statistics methods for analysis of multivariate data, focusing on data collected in form of repeated measurements. Multivariate normal distribution, Hotelling's T2, multivariate analysis of variance, repeated measures analysis of variance, growth curve models, mixed effects models. Methods for analyzing multivariate data in form of counts, categorical data and binary data, emphasizing recent approaches in statistical literature.
• ST 733 001 Applied Spatial Statistics (Montserrat Fuentes)
  Graphical and quantitative description of spatial data. Kriging, block kriging and cokriging. Common variogram models. Analysis of mean-nonstationary data by median polish and universal kriging. Spatial autoregressive models, estimation and testing. Spatial sampling procedures. Use of existing software with emphasis on analysis of real data from the environmental, geological and agricultural sciences.
• ST 744 001 Categorical Data Analysis (Daowen Zhang)
  Statistical models and methods for categorical responses including the analysis of contingency tables, logistic and Poisson regression, and generalized linear models. Survey of asymptotic and exact methods and their implementation using standard statistical software.
• ST 745 001 Analysis of Survival Data (Wenbin Lu)
  Statistical methods for analysis of time-to-event data, with application to situations with data subject to right-censoring and staggered entry, including clinical trials. Survival distribution and hazard rate; Kaplan-Meier estimator for survival distribution and Greenwood's formula; log-rank and weighted long-rank tests; design issues in clinical trials. Regression models, including accelerated failure time and proportional hazards; partial likelihood; diagnostics.
• ST/BMA/MA 772 001 Biomathematics II (Kevin Gross)
  Biomathematics II introduces stochastic models and their application to biological phenomena. Mathematical topics covered include discrete- and continuous-time Markov chains, single- and multi-type branching processes, Poisson processes, birth-death processes and basic diffusions. Biological applications include molecular motion, population genetics, DNA sequence evolution, and the dynamics of epidemic disease. Models will be studied through both analytical mathematics and computer simulation. Students will have the opportunity to engage the contemporary scientific literature with readings, discussions and small projects. Prerequisites: Students should be familiar with undergraduate-level probability theory. BMA / MA / ST 771 is not required.
• ST 782 002 Time Series Analysis: Time Domain (Donald Martin)
  Estimation inference for coefficients in autoregressive, moving average and mixed models and large sample. Distribution theory for autocovariances and their use in identification of time series models. Stationarity and seasonality. Extensions of theory and methods to multiple series including vector autoregressions, transfer function models, regression with time series errors, state space modelin
• ST 790 001 Theory of Data Mining (Helen Zhang); letter-graded
• ST 790 002 Bayesian Biostatistics (Sujit Ghosh); letter-graded
  (Pre-req ST 521) This course is an experimental offering focused on Bayesian inferential methods with emphasis on biostatistics applications. The essence of Bayesian methods is based on the concept of updating evidence using formal probabilistic rules. Unlike frequentist statistics, which attaches repeated-sampling frequencies to parameter estimators, Bayesian statistics directly describes uncertainty about unknown statistical parameters with a probability distribution. With this foundation, much of the Bayesian statistics follows from basic rules of probability theory and associated computational methods. In the past few years there have been a substantial change in attitudes of many biostatisticians and other applied statisticians toward implementation of the Bayesian paradigm. Recent developments of computational tools have brought Bayesian treatment of realistic, complex problems within the reach of practicing statisticians. Hence, knowledge of these techniques is critical for conducting research in most areas of contemporary biostatistics. This course will illustrate a variety of computational methods, simulation techniques, and hierarchical models suitable for analyzing clinical data. The primary emphasis will be on gaining an intuitive grasp of how the models work and what is needed to implement them for biostatistics research. Instead of spending much time on formal proofs, the course will use informal simulation experiments, case studies, and applied exercises to examine the intuition and working properties of the models. Using a wide variety of real data examples, this course will illustrate some key features of Bayesian inference, including the comparisons of means and proportions in the context of bioequivalence studies. A critical component of this course will be writing and running original codes using freely available software like R and WinBUGS.
• ST 810 001 SAMSI course (Montserrat Fuentes)
• ST 810 004 Preparation for Research (John Monahan)-1 credit hour; S/U grading