Multivariate Analysis

Most of the statistical methods we have studied so far treat variables one at a time, or treat one variable as dependent on a number of others (regression, ANOVA). There are clearly other possibilities. Two ideas that we pursue in this course are

Interdependence: a collection of related variables, with no dependent variable, and we try to understand the nature of the dependence.

Other types of dependence: typically with several related dependent variables, and we try to unravel the complicated patterns of dependence.

Depending on the goals of our analysis, we need to learn different ways of doing these things.

Course text is Analyzing Multivariate Data by James Lattin, Douglas Carroll and Paul Green, published by Duxbury. I like this book because it takes a solidly practical approach, using lots of words and pictures. We'll be covering Chapters 2 and 4-13, plus some extra bits and pieces (a reference for these is Applied Multivariate Statistical Analysis by Johnson and Wichern, published by Prentice-Hall -- be warned, this is not a very readable book!).

We'll be using SAS software for our analyses; I will try to show you what you need to know, and there will also be a manual available early in the term.

We meet for two lectures a week (Tuesday 12:00-13:00 in B361 and Thursday 11:00-12:00 in A332), plus one tutorial for which we will need to arrange a time. I will try to schedule this in the Math Computing Lab (on the 6th floor of the S wing).

How to find me:

by e-mail. I intend to respond to your e-mails within two working days.
by phone at 5654 (416-287-5654).

This is a small class, so I won't set formal office hours; if you want to see me, find me in my office or e-mail to agree a time. You might also find me this term in the Math Aid Room (S506F).