BootKNN function

Bootstrap with kNN

How to bootstrap with kNN (and DNN)

BootKNN(data, classes, sub="none", nsam=4, nboot=1000, misclass=TRUE, method="knn", ...)

Arguments

• data: Data frame to classify
• classes: Character vector of class names
• sub: Subsample to use (see example)
• nsam: Number of training items from each level of grouping factor, default 4
• nboot: Number of iterations
• misclass: Calculate misclassification table?
• method: Either "knn" (class::knn()) or "dnn" (shipunov::Dnn())
• ...: Further arguments to method functions

Details

This function samples equal numbers ('nsam') of training items from each level of grouping factor.

It also allows to use subset of data which will be used for sub-sampling of training data.

Returns

Returns all predictions as character matrix, each boot is a column

Author(s)

Alexey Shipunov

See Also

class::knn, Dnn

Examples

iris.sub <- 1:nrow(iris) %in% seq(1, nrow(iris), 5)
iris.bootknn <- BootKNN(iris[, -5], iris[, 5], sub=iris.sub)
## calculate and plot stability
st <- apply(iris.bootknn, 1, function(.x) var(as.numeric(as.factor(.x))))
plot(prcomp(iris[, -5])$x, col=iris$Species, pch=ifelse(st == 0, 19, 1))
## boot Dnn
BootKNN(iris[, -5], iris[, 5], nboot=50, method="dnn",
 k=1, FUN=function(.x) Gower.dist(.x))