% Generated by roxygen2: do not edit by hand
% Please edit documentation in R/rfe.R
\name{rfe}
\alias{rfe}
\alias{rfe.default}
\alias{rfeIter}
\alias{predict.rfe}
\alias{update.rfe}
\alias{rfe.formula}
\alias{rfe.recipe}
\title{Backwards Feature Selection}
\usage{
rfe(x, ...)

\method{rfe}{default}(
  x,
  y,
  sizes = 2^(2:4),
  metric = ifelse(is.factor(y), "Accuracy", "RMSE"),
  maximize = ifelse(metric \%in\% c("RMSE", "MAE", "logLoss"), FALSE, TRUE),
  rfeControl = rfeControl(),
  ...
)

\method{rfe}{formula}(form, data, ..., subset, na.action, contrasts = NULL)

rfeIter(
  x,
  y,
  testX,
  testY,
  sizes,
  rfeControl = rfeControl(),
  label = "",
  seeds = NA,
  ...
)

\method{update}{rfe}(object, x, y, size, ...)

\method{rfe}{recipe}(
  x,
  data,
  sizes = 2^(2:4),
  metric = NULL,
  maximize = NULL,
  rfeControl = rfeControl(),
  ...
)
}
\arguments{
\item{x}{A matrix or data frame of predictors for model training. This
object must have unique column names. For the recipes method, \code{x}
is a recipe object.}

\item{\dots}{options to pass to the model fitting function (ignored in
\code{predict.rfe})}

\item{y}{a vector of training set outcomes (either numeric or factor)}

\item{sizes}{a numeric vector of integers corresponding to the number of
features that should be retained}

\item{metric}{a string that specifies what summary metric will be used to
select the optimal model. By default, possible values are "RMSE" and
"Rsquared" for regression and "Accuracy" and "Kappa" for classification. If
custom performance metrics are used (via the \code{functions} argument in
\code{\link{rfeControl}}, the value of \code{metric} should match one of the
arguments.}

\item{maximize}{a logical: should the metric be maximized or minimized?}

\item{rfeControl}{a list of options, including functions for fitting and
prediction. The web page
\url{http://topepo.github.io/caret/recursive-feature-elimination.html#rfe} has more
details and examples related to this function.}

\item{form}{A formula of the form \code{y ~ x1 + x2 + ...}}

\item{data}{Data frame from which variables specified in
\code{formula} or \code{recipe} are preferentially to be taken.}

\item{subset}{An index vector specifying the cases to be used
in the training sample. (NOTE: If given, this argument must be
named.)}

\item{na.action}{A function to specify the action to be taken
if NAs are found. The default action is for the procedure to
fail. An alternative is \code{na.omit}, which leads to rejection
of cases with missing values on any required variable. (NOTE: If
given, this argument must be named.)}

\item{contrasts}{A list of contrasts to be used for some or all
the factors appearing as variables in the model formula.}

\item{testX}{a matrix or data frame of test set predictors. This must have
the same column names as \code{x}}

\item{testY}{a vector of test set outcomes}

\item{label}{an optional character string to be printed when in verbose
mode.}

\item{seeds}{an optional vector of integers for the size. The vector should
have length of \code{length(sizes)+1}}

\item{object}{an object of class \code{rfe}}

\item{size}{a single integers corresponding to the number of features that
should be retained in the updated model}
}
\value{
A list with elements \item{finalVariables}{a list of size
\code{length(sizes) + 1} containing the column names of the ``surviving''
predictors at each stage of selection. The first element corresponds to all
the predictors (i.e. \code{size = ncol(x)})} \item{pred }{a data frame with
columns for the test set outcome, the predicted outcome and the subset
size.}
}
\description{
A simple backwards selection, a.k.a. recursive feature elimination (RFE),
algorithm
}
\details{
More details on this function can be found at
\url{http://topepo.github.io/caret/recursive-feature-elimination.html}.

This function implements backwards selection of predictors based on
predictor importance ranking. The predictors are ranked and the less
important ones are sequentially eliminated prior to modeling. The goal is to
find a subset of predictors that can be used to produce an accurate model.
The web page \url{http://topepo.github.io/caret/recursive-feature-elimination.html#rfe}
has more details and examples related to this function.

\code{rfe} can be used with "explicit parallelism", where different
resamples (e.g. cross-validation group) can be split up and run on multiple
machines or processors. By default, \code{rfe} will use a single processor
on the host machine. As of version 4.99 of this package, the framework used
for parallel processing uses the \pkg{foreach} package. To run the resamples
in parallel, the code for \code{rfe} does not change; prior to the call to
\code{rfe}, a parallel backend is registered with \pkg{foreach} (see the
examples below).

\code{rfeIter} is the basic algorithm while \code{rfe} wraps these
operations inside of resampling. To avoid selection bias, it is better to
use the function \code{rfe} than \code{rfeIter}.

When updating a model, if the entire set of resamples were not saved using
\code{rfeControl(returnResamp = "final")}, the existing resamples are
removed with a warning.
}
\note{
We using a recipe as an input, there may be some subset
 sizes that are not well-replicated over resamples. `rfe` method
 will only consider subset sizes where at least half of the
 resamples have associated results in the search for an optimal
 subset size.
}
\examples{

\dontrun{
data(BloodBrain)

x <- scale(bbbDescr[,-nearZeroVar(bbbDescr)])
x <- x[, -findCorrelation(cor(x), .8)]
x <- as.data.frame(x, stringsAsFactors = TRUE)

set.seed(1)
lmProfile <- rfe(x, logBBB,
                 sizes = c(2:25, 30, 35, 40, 45, 50, 55, 60, 65),
                 rfeControl = rfeControl(functions = lmFuncs,
                                         number = 200))
set.seed(1)
lmProfile2 <- rfe(x, logBBB,
                 sizes = c(2:25, 30, 35, 40, 45, 50, 55, 60, 65),
                 rfeControl = rfeControl(functions = lmFuncs,
                                         rerank = TRUE,
                                         number = 200))

xyplot(lmProfile$results$RMSE + lmProfile2$results$RMSE  ~
       lmProfile$results$Variables,
       type = c("g", "p", "l"),
       auto.key = TRUE)

rfProfile <- rfe(x, logBBB,
                 sizes = c(2, 5, 10, 20),
                 rfeControl = rfeControl(functions = rfFuncs))

bagProfile <- rfe(x, logBBB,
                  sizes = c(2, 5, 10, 20),
                  rfeControl = rfeControl(functions = treebagFuncs))

set.seed(1)
svmProfile <- rfe(x, logBBB,
                  sizes = c(2, 5, 10, 20),
                  rfeControl = rfeControl(functions = caretFuncs,
                                          number = 200),
                  ## pass options to train()
                  method = "svmRadial")

## classification

data(mdrr)
mdrrDescr <- mdrrDescr[,-nearZeroVar(mdrrDescr)]
mdrrDescr <- mdrrDescr[, -findCorrelation(cor(mdrrDescr), .8)]

set.seed(1)
inTrain <- createDataPartition(mdrrClass, p = .75, list = FALSE)[,1]

train <- mdrrDescr[ inTrain, ]
test  <- mdrrDescr[-inTrain, ]
trainClass <- mdrrClass[ inTrain]
testClass  <- mdrrClass[-inTrain]

set.seed(2)
ldaProfile <- rfe(train, trainClass,
                  sizes = c(1:10, 15, 30),
                  rfeControl = rfeControl(functions = ldaFuncs, method = "cv"))
plot(ldaProfile, type = c("o", "g"))

postResample(predict(ldaProfile, test), testClass)

}

#######################################
## Parallel Processing Example via multicore

\dontrun{
library(doMC)

## Note: if the underlying model also uses foreach, the
## number of cores specified above will double (along with
## the memory requirements)
registerDoMC(cores = 2)

set.seed(1)
lmProfile <- rfe(x, logBBB,
                 sizes = c(2:25, 30, 35, 40, 45, 50, 55, 60, 65),
                 rfeControl = rfeControl(functions = lmFuncs,
                                         number = 200))


}


}
\seealso{
\code{\link{rfeControl}}
}
\author{
Max Kuhn
}
\keyword{models}