1\name{tune.control} 2\alias{tune.control} 3\title{Control Parameters for the Tune Function} 4\description{ 5 Creates an object of class \code{tune.control} to be used with 6 the \code{tune} function, containing various control parameters. 7} 8\usage{ 9tune.control(random = FALSE, nrepeat = 1, repeat.aggregate = mean, 10sampling = c("cross", "fix", "bootstrap"), sampling.aggregate = mean, 11sampling.dispersion = sd, 12cross = 10, fix = 2/3, nboot = 10, boot.size = 9/10, best.model = TRUE, 13performances = TRUE, error.fun = NULL) 14} 15\arguments{ 16 \item{random}{if an integer value is specified, \code{random} 17 parameter vectors are drawn from the parameter space.} 18 \item{nrepeat}{specifies how often training shall be repeated.} 19 \item{repeat.aggregate}{function for aggregating the repeated training results.} 20 \item{sampling}{sampling scheme. If \code{sampling = "cross"}, a 21 \code{cross}-times cross validation is performed. If \code{sampling 22 = "boot"}, \code{nboot} training sets of size \code{boot.size} (part) 23 are sampled (with replacement) from the supplied data. If \code{sampling 24 = "fix"}, a single split into training/validation set is 25 used, the training set containing a \code{fix} part of the supplied 26 data. Note that a separate validation set can be supplied via 27 \code{validation.x} and \code{validation.y}. It is only used for 28 \code{sampling = "boot"} and \code{sampling = "fix"}; in the latter 29 case, \code{fix} is set to 1.} 30 \item{sampling.aggregate,sampling.dispersion}{functions for aggregating the training 31 results on the generated training samples (default: mean and 32 standard deviation).} 33 \item{cross}{number of partitions for cross-validation.} 34 \item{fix}{part of the data used for training in fixed sampling.} 35 \item{nboot}{number of bootstrap replications.} 36 \item{boot.size}{size of the bootstrap samples.} 37 \item{best.model}{if \code{TRUE}, the best model is trained and 38 returned (the best parameter set is used for 39 training on the complete training set).} 40 \item{performances}{if \code{TRUE}, the performance results for all 41 parameter combinations are returned.} 42 \item{error.fun}{function returning the error measure to be minimized. 43 It takes two arguments: a vector of true values and a vector of 44 predicted values. If \code{NULL}, the misclassification error is used 45 for categorical predictions and the mean squared error for numeric 46 predictions.} 47} 48\value{ 49 An object of class \code{"tune.control"} containing all the above 50 parameters (either the defaults or the user specified values). 51} 52\author{ 53 David Meyer\cr 54 \email{David.Meyer@R-project.org} 55} 56\seealso{\code{\link{tune}}} 57\keyword{models} 58