multiple.ker.score.cv {RANKS}

RANKS multiple cross-validation for a single class

Description

Function to execute multiple cross-validation with RANKS for a single class.

Usage

multiple.ker.score.cv(RW, ind.pos, m = 5, p = 100, stratified=TRUE, 
                      init.seed = 0, fun = KNN.score, ...)

Arguments

RW

matrix. Kernel matrix or any valid symmetric matrix

ind.pos

indices of the positive examples. They are the row indices of RW corresponding to positive examples.

m

number of folds for each cross-validation

p

number of repeated cross-validations

stratified

boolean. If TRUE (def.) stratified CV is performed otherwise vanilla CV is done

init.seed

initial seed for the random generator (def: 0)

fun

function. It must be a kernel-based score method (default KNN.score)

...

optional arguments for the function fun

Details

It performs multiple cross-validation using RANKS to predict the cross-validated scores. The cross-validation is stratified: the folds are constructed separately for each class, to maintain an equal ratio between classes among folds. It computes the scores by averaging across multiple cross validations.

Value

A list with two components:

av.scores

a vector with the average scores across multiple cross-validations. Elements of the vector av.scores correspond to the rows of RW

pos.scores

a vector with the scores of positive elements collected at each iteration

See Also

ker.score.cv, multiple.ker.score.thresh.cv, rw.kernel-methods, Kernel functions.

Examples

# Nodel label ranking for the DrugBank category Penicillins
# on the Tanimoto chemical structure similarity network (1253 drugs)
# using 5 fold cross-validation repeated 10 times
# and eav-score with 1-step random walk kernel
library(bionetdata);
data(DD.chem.data);
data(DrugBank.Cat);
labels <- DrugBank.Cat[,"Penicillins"];
ind.pos <- which(labels==1);
K <- rw.kernel(DD.chem.data);
res <- multiple.ker.score.cv(K, ind.pos, m = 5, p = 10, init.seed = 0, fun = eav.score);
# the same but using the NN-score
res <- multiple.ker.score.cv(K, ind.pos, m = 5, p = 10, init.seed = 0, fun = NN.score);