R-help,
I'm trying to evaluate the misclasification rate
of an estimated discrimated function (hyDA)
hyDA - with(hyB, lda(Maturity ~ Length + Weight))
, so here is what I do (code from S-plus 6.1
Guide to Statistics Vol. 2 p 101)
test - predict(hyDA)
tbl - table(hyB$Maturity, test$class)
The error rates I get is 0.57 which is very large
I was wondering if I'm doing something wrong.
It is quite possible, if your groups are overlapping or not separated
by a linear boundary. Given that you have only two varaibles, have
you plotted them?
Hadley
