35 lines
1.3 KiB
Text
35 lines
1.3 KiB
Text
|
% Generated by roxygen2: do not edit by hand
|
||
|
|
% Please edit documentation in R/connectToData.R
|
||
|
|
\name{connectToData}
|
||
|
|
\alias{connectToData}
|
||
|
|
\title{Connect To Data}
|
||
|
|
\usage{
|
||
|
|
connectToData(forest, responses, covariateData)
|
||
|
|
}
|
||
|
|
\arguments{
|
||
|
|
\item{forest}{The forest to connect data too}
|
||
|
|
|
||
|
|
\item{responses}{The responses in the data; aka the left hand side of the formula}
|
||
|
|
|
||
|
|
\item{covariateData}{A data.frame containing all of the covariates used in the training dataset}
|
||
|
|
}
|
||
|
|
\value{
|
||
|
|
The same forest, but connected to the training data.
|
||
|
|
}
|
||
|
|
\description{
|
||
|
|
When a trained forest is saved, the training dataset is not saved alongside
|
||
|
|
it. When it's loaded back up, it can be more convenient (and in some cases
|
||
|
|
necessary) to import the training dataset back into the Java environment so
|
||
|
|
that it's readily accessible. There are only two functions that look for the
|
||
|
|
training dataset: \code{predict}, where you can easily just specify an
|
||
|
|
alternative dataset, or \code{\link{addTrees}}, which requires the training
|
||
|
|
dataset be connected.
|
||
|
|
}
|
||
|
|
\examples{
|
||
|
|
data <- data.frame(x1=rnorm(1000), x2=rnorm(1000), y=rnorm(1000))
|
||
|
|
forest <- train(y~x1+x2, data, ntree=100, numberOfSplits=0, nodeSize=1, mtry=1)
|
||
|
|
forest$dataset <- NULL # what the forest looks like after being loaded
|
||
|
|
|
||
|
|
forest <- connectToData(forest, data$y, data)
|
||
|
|
}
|