koplsCrossValSet {kopls} | R Documentation |
Generates sets of training/test observations useful for cross-validation (CV).
How the sets are generated is determined by the type
parameter,
which can be either 'nfold' for n-fold cross-validation, 'mccv' for Monte Carlo CV,
'mccvb' for Monte Carlo class-balanced CV.
koplsCrossValSet(K, Y, type = "nfold", nfold, i, trainFrac = (2/3))
K |
Kernel matrix. |
Y |
Response matrix. |
type |
Type of cross-validation: 'nfold' for n-fold, 'mccv' for Monte Carlo CV, 'mccvb' for Monte Carlo class-balanced CV. |
nfold |
Number of total nfold rounds (if type='nfold'). |
i |
Current nfold round (if type='nfold'). |
trainFrac |
Fraction of observations in training set. |
If type
is set to 'nfold', the parameter nfold
determines the number of rounds,
which are later subindexed by the i
parameter.
If 'mccv' or 'mccvb', the trainFrac
parameter determines the fraction of observations
that will belong to the training set; remaining observations end up in the test set.
List object with the following entries:
KTrTr |
Kernel training matrix; KTrTr = <phi(Xtr),phi(Xtr)>. |
KTeTr |
Kernel test/training matrix; KTeTr = <phi(Xte),phi(Xtr)>. |
KTeTe |
Kernel test matrix; KTeTe = <phi(Xte),phi(Xte)>. |
yTrain |
Y training set. |
yTest |
Y test set. |
trainInd |
Indices of training set observations. |
testInd |
Indices of test set observations. |
Max Bylesjo and Mattias Rantalainen
Rantalainen M, Bylesjo M, Cloarec O, Nicholson JK, Holmes E and Trygg J. Kernel-based orthogonal projections to latent structures (K-OPLS), J Chemometrics 2007; 21:376-385. doi:10.1002/cem.1071.