The R Project SVN R

\name{var}
\title{Covariance Matrices}
\usage{
var(x, y = x, na.rm = FALSE, use)
}
\alias{var}
\arguments{
\item{x}{a numeric matrix or vector.}
\item{y}{a numeric matrix or vector.}
\item{na.rm}{logical.}
\item{use}{an optional character string giving a
  method for computing covariances in the presence
  of missing values.  This must be one of \code{"all.obs"},
  \code{"complete.obs"} or \code{"pairwise.complete.obs"},
  with abbreviation being permitted.}
}
\description{
  \code{var} computes the variance of \code{x} and the
  covariance of \code{x} and \code{y} if \code{x} and \code{y}
  are vectors.  If \code{x} and \code{y} are matrices then
  the covariance between the columns of \code{x} and the
  the columns of \code{y} are computed.
}
\details{
  If \code{na.rm} is \code{TRUE} then the complete observations (rows)
  are used to compute the variance.  If \code{na.rm} is \code{FALSE}
  and there are missing values, then \code{var} will fail.

  The argument \code{use} can also be used for describing how
  to handle missing values.
  Specifying \code{use = "all"} is equivalent to specifying
  \code{na.rm = FALSE} and specifying \code{use = "pair"} is equivalent to
  \code{na.rm = TRUE}.
  If \code{use = "pair"}, then all the observations which are
  complete for a pair of variables are used to compute the
  covariance for that pair of variables.
  This can result in covariance matrices which are not
  positive semidefinite.
}
\seealso{
  \code{\link{cov}} with the same functionality for the
  multivariate case.
}
\examples{
var(1:10)# 9.166667

var(1:5,1:5)# 2.5
}
\keyword{univar}
\keyword{multivariate}