cv: Cross Validation

In ampel-leipzig/rusranger: Modified ranger implementation to support random-under-sampling

Cross Validation

Description

Runs a cross validation for train and prediction function.

Usage

cv(x, y, FUN, nfolds = 5, ...)

Arguments

x	matrix/data.frame, feature matrix, see ranger() for details.
y	numeric/factor, classification labels, see ranger() for details.
FUN	function function to optimize.
nfolds	integer(1) number of cross validation folds.
...	further arguments passed to FUN.

Value

double(1) median AUC across all cross validation splits

Note

The function to optimize has to accept five arguments: xtrain, ytrain, xtest, ytest and ....

Examples

.rusranger <- function(xtrain, ytrain, xtest, ytest, ...) {
    rngr <- rusranger(x = xtrain, y = ytrain, ...)
    pred <- as.numeric(predict(rngr, xtest)$predictions[, 2L])
    performance(prediction(pred, ytest), measure = "auc")@y.values[[1L]]
}
cv(iris[-5], as.numeric(iris$Species == "versicolor"), .rusranger, nfolds = 3)

ampel-leipzig/rusranger documentation built on April 28, 2022, 3:22 a.m.