SSC 385 Modern Statistic Methods

Instructor: Dr. Maggie Myers
Office: PAI 5.48 Phone: 471-9533
E-mail: myers@cs.utexas.edu
Office hours:

Topic

This course is intended to illustrate current approaches in modeling and computation in statistics.

Intended audience

Masters and Ph.D. students in machine learning, data mining, computational biology, engineering, psychology, and other fields in need of advanced modeling tools in their research.

Methods of evaluation

Homework, participation, midterm, and projects. Assignments will be made throughout the semester and will generally be due a week after assigned. A final project involves giving a brief presentation of some subtopic of interest.

Text (optional)

DeGroot, Morris H. and Mark J. Schervish, Probability and Statistics, 3rd ed. Addison Wesley, 2002. The text is a reference and will be supplemented. Information will be organized on the classÍs electronic blackboard (http://courses.utexas.edu)

Syllabus

Background
- Probability and Bayes Rule
- Random Variables and Vectors: Discrete and Continuous
- Using Bayesian Networks
- Expectation, Variance, and the Central Limit Theorem
Inference
- Estimation and Distributions of Estimators
- n Bayesian Modeling, Prior and Posterior Distributions
- MLE/MAP/EM Parameter Estimation
- Testing Hypotheses
Models
- Linear Statistical Models
- Mixture Models
- Markov Models / Random Walks
- Hidden Markov Models
Simulations
- Bootstrapping
- Markov Chain Monte Carlo Methods
- Randomized Sampling Algorithms (Monte Carlo)
- Metropolis Hastings Algorithms
- Gibbs Sampling

These concepts will be illustrated using applications drawn from machine learning, computational biology, and other interests of the students enrolled. Students will apply techniques to data sets using R and other appropriate packages.