I’ve taught a variety of courses over the last 20 years, ranging from freshman level introductory courses to PhD theory courses for Statistics majors to topics courses for graduate students across the majors. I enjoy incorporating student research and research with collaborators into the curriculum whenever possible. I particularly enjoy final projects, where students from a variety of backgrounds apply what they’ve learned and share their findings with the class.
Current Teaching at OSU
-
Statistical Machine Learning with R STAT 4463 / 5063. This is an applied statistical learning course for statistics majors and graduate students in other quantitative areas that have some statistics background - ideally an applied regression analysis course that uses R. We cover most topics in the popular book Applied Statistical Learning with Application in R by Gareth, Witten, Hastie and Tibshirani. Supervised learning methods are: review of multiple linear and logistic regression models, k nearest neighbors, disciminant functions, penalized regression methods such as LASSO and ridge regression, principle component regression and partial least squares, other nonlinear methods such as local regression, natural splines, tree based methods and generalized additive models. Unsupervised methods include principle component analysis and cluster analysis. Modell selection topics/methods include information measures like AIC, BIC, bias-variance tradeoff, overfitting, prediction error, cross validation and bootstrapping. Upon completion of the course, students will be able to assess, compare, and apply modern machine learning methods to learn from their data.
-
R Programming STAT 4191/5191/4193/5193. This is an introductory level course in R programming for statistics BS/MS students and students in other quantitative areas that have at least one senior level statistics course. We cover core data structures (matrices, data frames, lists, factors), plotting, functions, loops, logicals and processing methods like random sampling and subsetting. Some popular packages, like ggplot, are covered too. But the focus is on in-depth understanding of R that leaves students capable of learning new functions, packages, and processing methods they may need to analyze their data.
-
Statistical Computing STAT 5093. This is a theory course focusing on algorithms for optimization and integration arising in Statistics. It is designed for PhD Statistics students and graduate students in other majors with at least two graduate level courses in mathematical statistics and vector calculus. Optimization algorithms include a review of Newton-Rhapson and gradient decent methods, Gauss-Newtwon methods for nonlinear least squares, Fisher Scoring and iteratively reweighted least squares algorithms for generalized linear models, proximal gradient decent and cyclclical coordinant decent methods for LASSO and other penalized regression problems, and the EM algorithm for mixture models and missing data. Integration methods include Gibbs, Metropolis-Hasting’s, and Markov Chain Monte Carlo methods with applications to Bayesian inference, missing data. Bootstrapping, cross validation, jacknife, and exact tests are studied.
-
Advanced Probability Theory STAT 6113. This a measure theoretic treatment of probablity theory for PhD students in Statistics or Mathematics. Topics include the construction of a probability measure, expectation operator, inequalities in statistics, modes of convergence, Martingales and other topics in stochastic processes.
-
Elements of Statistics STAT 2013. This is an introductory level online course for undergraduate social science majors. It covers descriptive statistics and statistical inference for one and two samples.
Past Teaching
- Engineering Statistics
- Elements of Statistics
- Business Statistics
- Multivariate Methods
- Multivariate Theory
- Time Series Analysis
- Statistical Inference