Multivariate Analysis II

Winter/Spring 2009; MW 10:30-12:00; THR 3079


The aim of this course is to introduce students to classical and modern multivariate techniques, such as principal components, factor and discriminant analyses. This course will treat both theory and data and emphasize troublesome issues encountered in practice. Students should be able to apply the core methods fluently by the end of the course.

Although algebra is necessarily an important part of most statistical training, Multivariate Analysis is one of the most pleasingly geometric subjects within statistics, where your intuition can be a reliable guide. This will not be a mathematical statistics course; rather this course will develop intuition for how to use multivariate methods effectively, and develop your judgment to choose which methods are most appropriate for particular problems. As medicine becomes increasingly driven by multiple measures recorded on individuals, this body of knowledge will assume a larger role.

Text: Johnson & Wichern: Applied Multivariate Statistical Analysis
Supplementary Text: Hastie, Tibshirani, and Friedman: The Elements of Statistical Learning



  1. Multivariate Regression
  2. Principal Components
  3. Factor Analysis & SEM
  4. Canonical Correlations
  5. Nonnegative Matrix Factorization
  6. Independent Components Analysis
  7. Robust multivariate methods
  8. Discriminant analysis
  9. Support vector machines
  10. Clustering
  11. Multidimensional scaling

Data and Scripts

8-4 Class Demos