Cross-Validation for fastknn

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

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.

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
## ---------------------------------------------