Phone: (416)-736-5250

Taught by: Professor Gene Denzel

Office: N615 Ross

Office Hours: M,W 9-10 and M 3:30-4:30; or by appointment

Web page: http://www.yorku.ca/lezned

 This web page will point you to the course pages, where details of assignments, programming references, other resources, etc will be provided. This site should be checked regularly.

Textbook: Applied Multivariate Statistical Analysis (4^th Edition) (Prentice-Hall 1998)

General outline: This course will deal mostly with the classical treatment of multivariate data, based on the multivariate normal distribution and its properties, albeit from a modern perspective. There is a wealth of material to study, even under that rubric. In recent years a large number of new techniques have been developed, making use of the substantial improvements in computing power now available. In this course, we will only be able to point out from time to time where such newer developments might be relevant, as areas for later study for those interested. Our approach, following the textbook, will be based on an emphasis on understanding how to apply these techniques. That will require, however, a thorough understanding of the underlying linear algebra and geometry which has made the multivariate normal distribution so powerful a tool.

The material to be covered is basically that in chapters 1-4 as basic foundation, most of the topics in chapters 5-7, plus a selection of topics from chapters 8-12 as time and interest allows. Students who have particular interest in any of the topics in the latter chapters should let me know, so that we can try to incorporate them into the course.

It is essential to understand that one learns statistics by doing, not by just reading. In order to be successful in this course it will be necessary to keep up with assignments on a weekly basis. To do the assignments will usually require use of software. Much of what we will be doing can be done within the SAS system, and all of it can be done with SPLUS, especially the graphical displays which will be so important for some of the methods. Both packages are available through the Math Department servers via the Gauss Lab (S110), or for graduate students the N604 lab. They are also available on phoenix, the central academic computing cluster. Students should ensure through MAYA that they have the appropriate accounts.

Grading: The grading in the course will be based on the following breakdown.

Final exam 35%

Midterm: 25%

Assignments: 20%

Project: 25% The project will be, for undergraduate students, a major data-analysis using many of the techniques to be studied in this course (plus, perhaps, others as appropriate). Graduate students will be expected to do more, and may be able to combine this with a programming project in SPLUS. More details will be available later.

Group work is encouraged, although individual writeups of the assignments are to be done. The project may be done in groups of two if desired, but we will need to discuss individually the terms for such arrangements.

The letter grade in the course will be based on the combination of the above grades, converted to a letter in rough correspondence to the common York mapping for undergraduates, but the exact break points will be determined by looking at components of the grade for those near the borderlines. Thus, a combined mark of 70.0 will not necessarily result in a grade of B, nor will a mark of 69.9 necessarily result in a C+. Graduate students will be marked on the usual graduate scale.