Cheg 258 - Lecture Notes - Mar. 2, 2001

Announcements

• An additional source of information on resampling methods (e.g., undersampling and the bootstrap) is found on the resampling statistics home page. Note that they are trying to sell you their packages - While I don't endorse their products (I've never looked at them) they do provide you with useful supplementary information on the technique.

Class notes

The main points of the lecture were

• Undersampling and the bootstrap

Goals:

• Two methods of estimating error, capable of handling non-linear models.

Reading

• The class notes.
• Supplementary (optional) reading from the resampling statistics home page. The introductory text from the University of Maryland is useful.

Question of the Day

• Given the same set of N measurements, which method is faster: undersampling or bootstrapping?

Matlab Example Codes

example 18a

example 18b