These are the data used in Figures 2.1-2.3, and elsewhere through the
book. The data are in the format of an Splus data.dump (ascii file); 
to restore the data as a list in Splus, use 

data.restore("mixture.example.data")

and you will now have an object called mixture.example.data. 
The reason for this format is that the objects are all of
different sizes. For non-Splus users, the format of this file is
not too hard to deparse, given the size information below.

The components are:

x	       200 x 2 matrix of training predictors
y	       class variable; logical vector of TRUES and 
	       FALSES - 100 of each
xnew	       matrix 6831 x 2 of lattice points in predictor space
prob	       vector of 6831 probabilities (of class TRUE) at each 
	       lattice point
marginal       marginal probability at each lattice point
px1	       69 lattice coordinates for x.1
px2	       99 lattice values for x.2  (69*99=6831)
means	       20 x 2 matrix of the mixture centers, first ten for one
	       class, next ten for the other

So for example, the Bayes error rate is computed as
bayes.error<-sum(marginal*(prob*I(prob<0.5)+(1-prob)*I(prob>=.5)))

If pred is a vector of predictions (of the logit, say):

pred<-predict.logit(xnew)

then the test error is 

test.error<-sum(marginal*(prob*I(pred <0)+(1-prob)*I(pred>=0)))