Kernel Principal Components Analysis

Performs a kernel PCA.

kernel.pca(K, ncomp = nrow(K$kernel))

Arguments

K

a kernel object obtained using either compute.kernel or combine.kernels.

ncomp

integer. Indicates the number of components to return..

Value

kernel.pca returns an object of classes "kernel.pca"

and "pca", which is a list containing the following entries:

ncomp

: the number of principal components;

X

: the input kernel matrix;

kernel

: the input kernel object provided by the user;

sdev

: the singular values (square root of the eigenvalues);

rotation

: the matrix of variable loadings (i.e., a matrix whose columns contain the eigenvectors);

loadings

: same as 'rotation' to keep the mixOmics spirit;

x

: same as 'rotation' to keep the mixOmics spirit;

References

Scholkopf B., Smola A. and Muller K.R. (1998) Nonlinear component analysis as a kernel eigenvalue problem. Neural Computation, 10, 1299-1319.

Author

Jerome Mariette <jerome.mariette@inrae.fr> Nathalie Vialaneix <nathalie.vialaneix@inrae.fr>

Examples

data(TARAoceans)
phychem.kernel <- compute.kernel(TARAoceans$phychem, kernel.func = "linear")
kernel.pca.result <- kernel.pca(phychem.kernel, ncomp = 3)