61.182 Informaatiotekn. ek II, spring 1997

Exercise problems

(Ch. 2, Ripley)

5. Ripley says (p. 67) that the error count, say *R*, has a binomial
distribution with mean *ME* and variance *ME*(1-*E*) where *E* is the
(unknown) probability of error and *M* is the size of the test set.
Show that this is true.

(Ch. 3, Ripley)

6. When is the classical LDA classifier (Ch. 3.1) the same as the classifier based on Fisher's linear discriminant?

Fri Apr 11 10:26:16 DST 1997