This function generates two datasets according to the model [AkBkQkDk] of the HDDA gaussian mixture model paramatrisation (see ref.).
Usage
simuldata(nlearn, ntest, p, K = 3, prop = NULL, d = NULL, a = NULL, b = NULL)
Value
X
The learning dataset.
clx
The class vector of the learning dataset.
Y
The test dataset.
cly
The class vector of the test dataset.
prms
The principal parameters used to generate the datasets.
Arguments
nlearn
The size of the learning dataset to be generated.
ntest
The size of the testing dataset to be generated.
p
The number of variables.
K
The number of classes.
prop
The proportion of each class.
d
The dimension of the intrinsic subspace of each class.
a
The value of the main parameter of each class.
b
The noise of each class.
Author
Laurent Berge, Charles Bouveyron and Stephane Girard
References
Bouveyron, C. Girard, S. and Schmid, C. (2007) “High Dimensional Discriminant Analysis”, Communications in Statistics : Theory and Methods, vol. 36(14), pp. 2607--2623
data <- simuldata(500, 1000, 50, K=5, prop=c(0.2,0.25,0.25,0.15,0.15))
X <- data$X
clx <- data$clx
f <- hdda(X, clx)
Y <- data$Y
cly <- data$cly
e <- predict(f, Y, cly)