MATH 742/842 Spring 2006

Multivariate Statistical Methods

Section 1: MW 12:10 - 2:00pm Distance Learning room (MUB ground floor)

Instructor: Philip J Ramsey, Ph.D., Trailer in C Lot (office T14)

Office Hours: MW 2:10-4:00 pm, or by appointment

Home Phone: (603)672-5651; Office: (603)862-2613

Far View: The course may be taken in person in the Distance Learning room or live over the Internet via using the Learnlinc software (free to enrollees). The Internet environment is fully interactive with the instructor. All class sessions are recorded and the recordings are available to students throughout the semester.

Textbook: Applied Multivariate Data Analysis 2^nd edition (paperback), Everitt and Dunn (2001), Arnold/Oxford University Press. The Chicago Guide to Writing About Multivariate Statistics (paperback), Jane E. Miller (2005), University of Chicago Press.

Purpose: To provide a one-semester course in multivariate statistical methods. Math 842 designation is typically for graduate students who have at least a bachelors degree. Undergraduates enroll in Math 742.

Computer Usage: JMP-IN 5.1 (student version) or JMP 6.0 professional statistical software is integrated into the course. The software has been installed on the university PC - and Macintosh networks of computers. Copies of JMP-IN 5.1 can be obtained through the Durham Book Exchange for around $80.00 (an excellent buy!). In addition some use may be made of the R statistical software.

Homework: Formal homework assignments will be collected and graded.

Project: One or more project reports involving the analysis of multivariate datasets are required and will constitute a significant portion of the grade. A take home exam may be given in lieu of a project.

Final Grade: The final grade is based upon the final project report, completion of the homework, and class participation. Approximately 60% of the grade will based on the homework and 40% on the projects.

Web Page: I will make available copies of the course slides and other materials, on the UNH Blackboard system. All students enrolled in the course automatically receive a Blackboard account.

Tentative Outline of Course Topics

(not all of the topics may be covered depending upon time)

Introduction to multivariate statistical concepts and matrix notation

Graphical Methods for Multivariate Data and Outlier Analysis

Principal Components Analysis

Factor Analysis

Review of Multiple Linear Regression

Partial Least Squares Regression

Multivariate Logistic Regression

Correspondence Analysis

Generalized Linear Models

Discriminant Analysis

Canonical Correlation Analysis

MANOVA

Cluster Analysis

Multidimensional Scaling

Data Mining (neural nets and CART models)