fastknnCV: Cross-Validation for fastknn

In davpinto/fastknn: Build Fast k-Nearest Neighbor Classifiers

Description

Does n-fold cross-validation for fastknn to find the best k parameter.

Usage

fastknnCV(x, y, k = 3:15, method = "dist", normalize = NULL, folds = 5,
  eval.metric = "overall_error", nthread = 1)

Arguments

x	input matrix of dimension nobs x nvars.
y	factor array wtih class labels for the x rows.
k	sequence of possible k values to be evaluated (default is [3:15]).
method	the probability estimator as in fastknn.
normalize	variable scaler as in fastknn.
folds	number of folds (default is 5) or an array with fold ids between 1 and n identifying what fold each observation is in. The smallest value allowable is nfolds=3. The fold assigment given by fastknnCV does stratified sampling.
eval.metric	classification loss measure to use in cross-validation. See classLoss for more details.
nthread	the number of CPU threads to use (default is 1).

Value

list with cross-validation results:

• best_eval: the best loss measure found in the cross-validation procedure.

• best_k: the best k value found in the cross-validation procedure.

• cv_table: data.frame with the test performances for each k on each data fold.

Author(s)

David Pinto.

See Also

classLoss

Examples

## Not run: 
library("mlbench")
library("caTools")
library("fastknn")

data("Ionosphere")

x <- data.matrix(subset(Ionosphere, select = -Class))
y <- Ionosphere$Class

set.seed(1024)
tr.idx <- which(sample.split(Y = y, SplitRatio = 0.7))
x.tr <- x[tr.idx,]
x.te <- x[-tr.idx,]
y.tr <- y[tr.idx]
y.te <- y[-tr.idx]

set.seed(2048)
cv.out <- fastknnCV(x = x.tr, y = y.tr, k = c(5,10,15,20), eval.metric="logloss")

cv.out$cv_table

## End(Not run)

davpinto/fastknn documentation built on May 15, 2019, 1:18 a.m.