Seminar on HTF2

Summer & Fall 2009

Week 1 : May 29

• Link to authors' web site for book
• Link to additional errata (besides what is given on the authors' web site).
• Link to home page for R
• R code for Sec. 2.3 example and extensions (Code was developed using version 2.8.1 of R. 2.9.0 must be different with regard to random numbers and will produce results that don't match my comments. 2.1.0 results are almost the same as 2.8.1.)

Week 2 : June 5

• Introduction to LDA
• Introduction to QDA and RDA
• R code for vowel data considered in Ch. 4, dealing with just two classes
• Training set for vowel data considered in Ch. 4 (copied from authors' web site)
• Test set for vowel data considered in Ch. 4 (copied from authors' web site)

Week 3 : June 12

• R code for vowel data considered in Ch. 4, dealing with all 11 classes
• more R code for vowel data considered in Ch. 4, dealing with all 11 classes (featuring canonical variables and further exploration of the differences between the test data and the training data)

Week 4 : June 19

• Introduction to logistic regression
• more R code for vowel data considered in Ch. 4, dealing with all 11 classes (featuring using weight functions and alternative metrics for KNN)
• more R code for vowel data considered in Ch. 4, dealing with both just two classes and also all 11 classes (featuring using logistic regression for classification)

Week 5 : June 26

• more R code for vowel data considered in Ch. 4, dealing with all 11 classes (featuring using methods from Sec. 13.2)
• R code extension of Sec. 2.3 example (featuring using methods from Sec. 13.2)

July 3

• No meeting this week (due to university being closed)

Week 6 : July 10

• R code extension of Sec. 2.3 example (featuring using methods from Sec. 13.2, with the main new feature being R functions for cross-validation routines for the Sec. 13.2 methods)
• R code for a comparison of the knn and kknn functions
• R code for solution of Exercise 13.7 ("easy" problem only) on p. 483 of HTF2
• R code for solution of Exercise 13.7 ("hard" problem only) on p. 483 of HTF2

Week 7 : July 17

• Introduction to Separating Hyperplanes and Support Vector Machines (from a course I taught in Summer 2005, but updated to correspond to the 2nd edition of HTF)
• Dr. Griva's presentation on Support Vector Machines

Week 8 : July 24

• R code for example of using support vector machines (in a setting where they do very well)

Week 9 : July 31

• R code for example of using the global dimension reduction method of subsection 13.4.2 (in a setting where it doesn't work so well compared to other methods)
• R code for example of using the global dimension reduction method of subsection 13.4.2 (in a setting where it works better than ordinary KNN), and also using support vector machines (although the SVMs don't work well with this data)

Week 10 : September 4

• R code for example of using the DANN method of Section 13.4

Week 11 : September 11

• Training set predictor values for gene expression data considered in Sec. 18.2 (copied from authors' web site)
• Training set response values for gene expression data considered in Sec. 18.2 (copied from authors' web site)
• Test set predictor values for gene expression data considered in Sec. 18.2 (copied from authors' web site)
• Test set response values for gene expression data considered in Sec. 18.2 (copied from authors' web site)
• R code for DDA and NSC methods of Section 18.2

Week 12 : September 18

• updated R code for gene expression data used 9/11
• Jill's presentation on clustering
• Jill's R code for her clustering example

• R code for investigating something puzzling about the svm function

January 29

• R code for variable selection when doing KNN
• R code for generating data to use with classification methods

April 2

• R code for a DANN example (where DANN works well)

June 18

• R code (and assorted comments) for an overview of classification methods
• R code for a 2nd example of using various classification methods
• R code for a 3rd example of using various classification methods

June 25

• R code for variable selection/weighting methods
• more R code for variable selection/weighting methods