The R Project SVN R

\name{predict.glm}
\title{Predict Method for GLM Fits}
\usage{
predict.glm(object, newdata = NULL, type = c("link", "response", "terms"),
            se.fit = FALSE, dispersion = NULL, terms = NULL, \dots)
}
\alias{predict.glm}
\arguments{
 \item{object}{a fitted object of class inheriting from \code{"glm"}.}
 \item{newdata}{optionally, a new data frame from which to make the
   predictions.  If omitted, the fitted linear predictors are used.}
 \item{type}{the type of prediction required.  The default is on the
   scale of the linear predictors; the alternative \code{"response"}
   is on the scale of the response variable.  Thus for a default
   binomial model the default predictions are of log-odds (probabilities
   on logit scale) and \code{type = "response"} gives the predicted
   probabilities.  The \code{"terms"} option returns a matrix giving the
   fitted values of each term in the model formula on the linear predictor
   scale.

   The value of this argument can be abbreviated.
 }
 \item{se.fit}{logical switch indicating if standard errors are required.}
 \item{dispersion}{the dispersion of the GLM fit to be assumed in
   computing the standard errors.  If omitted, that returned by
   \code{summary} applied to the object is used.}
  \item{terms}{with \code{type="terms"} by default all terms are returned.
   A character vector specifies which terms are to be returned}
}
\description{
  Obtains predictions and optionally estimates standard errors of those
  predictions from a fitted generalized linear model object.
}
\value{
  If \code{se = FALSE}, a vector or matrix of predictions.  If \code{se
    = TRUE}, a list with components
  \item{fit}{Predictions}
  \item{se.fit}{Estimated standard errors}
  \item{residual.scale}{A scalar giving the square root of the
    dispersion used in computing the standard errors.}
}
\author{B.D. Ripley}
%  \note{This method is also currently used for objects of
%    class \code{"survreg"} (parametric survival fits from package
%    \code{survival4}) and possibly others. The assumptions made by
%    \code{predict.glm} may not always be right for such objects.}

\seealso{\code{\link{glm}}}

\examples{
## example from Venables and Ripley (1997, pp. 231-3.)
ldose <- rep(0:5, 2)
numdead <- c(1, 4, 9, 13, 18, 20, 0, 2, 6, 10, 12, 16)
sex <- factor(rep(c("M", "F"), c(6, 6)))
SF <- cbind(numdead, numalive=20-numdead)
budworm.lg <- glm(SF ~ sex*ldose, family=binomial)
summary(budworm.lg)

plot(c(1,32), c(0,1), type="n", xlab="dose",
   ylab="prob", log="x")
text(2^ldose, numdead/20,as.character(sex))
ld <- seq(0, 5, 0.1)
lines(2^ld, predict(budworm.lg, data.frame(ldose=ld,
   sex=factor(rep("M", length(ld)), levels=levels(sex))),
   type="response"))
lines(2^ld, predict(budworm.lg, data.frame(ldose=ld,
   sex=factor(rep("F", length(ld)), levels=levels(sex))),
   type="response"))
}
\keyword{models}
\keyword{regression}