widekernelpls.fit.rd

做主成分回归和偏最小二乘回归

%% $Id: widekernelpls.fit.Rd 145 2007-10-12 08:56:32Z bhm $
\encoding{latin1}
\name{widekernelpls.fit}
\alias{widekernelpls.fit}
\title{Wide Kernel PLS (R鋘nar et al.)}
\description{Fits a PLSR model with the wide kernel algorithm.}
\usage{widekernelpls.fit(X, Y, ncomp, stripped = FALSE,
                  tol = .Machine$double.eps^0.5, maxit = 100, \dots)}
\arguments{
  \item{X}{a matrix of observations.  \code{NA}s and \code{Inf}s are not
    allowed.} 
  \item{Y}{a vector or matrix of responses.  \code{NA}s and \code{Inf}s
    are not allowed.} 
  \item{ncomp}{the number of components to be used in the
    modelling.}
  \item{stripped}{logical.  If \code{TRUE} the calculations are stripped
    as much as possible for speed; this is meant for use with
    cross-validation or simulations when only the coefficients are
    needed.  Defaults to \code{FALSE}.}
  \item{tol}{numeric.  The tolerance used for determining convergence in
    the algorithm.}
  \item{maxit}{positive integer.  The maximal number of iterations used
    in the internal Eigenvector calculation.}
  \item{\dots}{other arguments.  Currently ignored.}
}
\details{This function should not be called directly, but through
  the generic functions \code{plsr} or \code{mvr} with the argument
  \code{method="widekernelpls"}.  The wide kernel PLS algorithm is
  efficient when the number of variables is (much) larger
  than the number of observations.  For very wide \code{X}, for instance
  12x18000, it can be twice as fast as \code{\link{kernelpls.fit}} and
  \code{\link{simpls.fit}}.  For other matrices, however, it can be much
  slower.  The results are equal to the results of the NIPALS algorithm.
}
\value{A list containing the following components is returned:
  \item{coefficients}{an array of regression coefficients for 1, \ldots,
    \code{ncomp} components.  The dimensions of \code{coefficients} are
    \code{c(nvar, npred, ncomp)} with \code{nvar} the number
    of \code{X} variables and \code{npred} the number of variables to be
    predicted in \code{Y}.}
  \item{scores}{a matrix of scores.}
  \item{loadings}{a matrix of loadings.}
  \item{loading.weights}{a matrix of loading weights.}
  \item{Yscores}{a matrix of Y-scores.}
  \item{Yloadings}{a matrix of Y-loadings.}
  \item{projection}{the projection matrix used to convert X to scores.}
  \item{Xmeans}{a vector of means of the X variables.}
  \item{Ymeans}{a vector of means of the Y variables.}
  \item{fitted.values}{an array of fitted values.  The dimensions of
    \code{fitted.values} are \code{c(nobj, npred, ncomp)} with
    \code{nobj} the number samples and \code{npred} the number of
    Y variables.}
  \item{residuals}{an array of regression residuals.  It has the same
    dimensions as \code{fitted.values}.}
  \item{Xvar}{a vector with the amount of X-variance explained by each
    number of components.}
  \item{Xtotvar}{Total variance in \code{X}.}

  If \code{stripped} is \code{TRUE}, only the components
  \code{coefficients}, \code{Xmeans} and \code{Ymeans} are returned.
}
\note{
  The current implementation has not undergone extensive testing yet,
  and should perhaps be regarded as experimental.  Specifically, the
  internal Eigenvector calculation does not always converge in extreme
  cases where the Eigenvalue is close to zero.  However, when it does
  converge, it always converges to the same results as
  \code{\link{kernelpls.fit}}, up to numerical inacurracies.

  The algorithm also has a bit of overhead, so when the number of
  observations is moderately high, \code{\link{kernelpls.fit}} can be
  faster even if the number of predictors is much higher.  The relative
  speed of the algorithms can also depend greatly on which BLAS and/or
  LAPACK library \R is linked against.
}
\references{
  R鋘nar, S., Lindgren, F., Geladi, P. and Wold, S. (1994) A PLS
  Kernel Algorithm for Data Sets with Many Variables and Fewer
  Objects.  Part 1: Theory and Algorithm.
  \emph{Journal of Chemometrics}, \bold{8}, 111--125.
}
\author{Bj鴕n-Helge Mevik}
\seealso{
  \code{\link{mvr}}
  \code{\link{plsr}}
  \code{\link{pcr}}
  \code{\link{kernelpls.fit}}
  \code{\link{simpls.fit}}
  \code{\link{oscorespls.fit}}
}
\keyword{regression}
\keyword{multivariate}