\name{gssanova0}
\alias{gssanova0}
\title{Fitting Smoothing Spline ANOVA Models with Non-Gaussian Responses}
\description{
    Fit smoothing spline ANOVA models in non-Gaussian regression.  The
    symbolic model specification via \code{formula} follows the same
    rules as in \code{\link{lm}} and \code{\link{glm}}.
}
\usage{
gssanova0(formula, family, type=NULL, data=list(), weights, subset,
          offset, na.action=na.omit, partial=NULL, method=NULL,
          varht=1, nu=NULL, prec=1e-7, maxiter=30)
}
\arguments{
    \item{formula}{Symbolic description of the model to be fit.}
    \item{family}{Description of the error distribution.  Supported
	are exponential families \code{"binomial"}, \code{"poisson"},
	\code{"Gamma"}, \code{"inverse.gaussian"}, and
	\code{"nbinomial"}.  Also supported are accelerated life model
	families \code{"weibull"}, \code{"lognorm"}, and
	\code{"loglogis"}.}
    \item{type}{List specifying the type of spline for each variable.
        See \code{\link{mkterm}} for details.}
    \item{data}{Optional data frame containing the variables in the
	model.}
    \item{weights}{Optional vector of weights to be used in the
	fitting process.}
    \item{subset}{Optional vector specifying a subset of observations
	to be used in the fitting process.}
    \item{offset}{Optional offset term with known parameter 1.}
    \item{na.action}{Function which indicates what should happen when
	the data contain NAs.}
    \item{partial}{Optional extra unpenalized terms in partial spline
	models.}
    \item{method}{Score used to drive the performance-oriented
	iteration.  Supported are \code{method="v"} for GCV,
	\code{method="m"} for GML, and \code{method="u"} for Mallow's CL.}
    \item{varht}{Dispersion parameter needed for \code{method="u"}.
	Ignored when \code{method="v"} or \code{method="m"} are
	specified.}
    \item{nu}{Inverse scale parameter in accelerated life model
        families.  Ignored for exponential families.}
    \item{prec}{Precision requirement for the iterations.}
    \item{maxiter}{Maximum number of iterations allowed for
	performance-oriented iteration, and for inner-loop multiple
	smoothing parameter selection when applicable.}
}
\details{
    The model specification via \code{formula} is intuitive.  For
    example, \code{y~x1*x2} yields a model of the form
    \deqn{
	y = C + f_{1}(x1) + f_{2}(x2) + f_{12}(x1,x2) + e
    }
    with the terms denoted by \code{"1"}, \code{"x1"}, \code{"x2"}, and
    \code{"x1:x2"}.

    The model terms are sums of unpenalized and penalized
    terms. Attached to every penalized term there is a smoothing
    parameter, and the model complexity is largely determined by the
    number of smoothing parameters.

    Only one link is implemented for each \code{family}.  It is the
    logit link for \code{"binomial"}, and the log link for
    \code{"poisson"}, \code{"Gamma"}, and \code{"inverse.gaussian"}.
    For \code{"nbinomial"}, the working parameter is the logit of the
    probability \eqn{p}; see \code{\link{NegBinomial}}.  For
    \code{"weibull"}, \code{"lognorm"}, and \code{"loglogis"}, it is the
    location parameter for the log lifetime.

    The models are fitted by penalized likelihood method through the
    performance-oriented iteration as described in the reference; the
    \eqn{O(n^{3})} algorithms of RKPACK are used for numerical
    calculations.  For \code{family="binomial"}, \code{"poisson"},
    \code{"nbinomial"}, \code{"weibull"}, \code{"lognorm"}, and
    \code{"loglogis"}, the score driving the performance-oriented
    iteration defaults to \code{method="u"} with \code{varht=1}.  For
    \code{family="Gamma"} and \code{"inverse.gaussian"}, the default is
    \code{method="v"}.
}
\section{Responses}{
    For \code{family="binomial"}, the response can be specified either
    as two columns of counts or as a column of sample proportions plus a
    column of total counts entered through the argument \code{weights},
    as in \code{\link{glm}}.

    For \code{family="nbinomial"}, the response may be specified as two
    columns with the second being the known sizes, or simply as a single
    column with the common unknown size to be estimated through the
    maximum likelihood.

    For \code{family="weibull"}, \code{"lognorm"}, or \code{"loglogis"},
    the response consists of three columns, with the first giving the
    follow-up time, the second the censoring status, and the third the
    left-truncation time.  For data with no truncation, the third column
    can be omitted.
}
\value{
    \code{gssanova0} returns a list object of \code{class}
    \code{c("gssanova0","ssanova0","gssanova")}.

    The method \code{\link{summary.gssanova0}} can be used to obtain
    summaries of the fits.  The method \code{\link{predict.ssanova0}}
    can be used to evaluate the fits at arbitrary points along with
    standard errors, on the link scale.  The methods
    \code{\link{residuals.gssanova}} and \code{\link{fitted.gssanova}}
    extract the respective traits from the fits.
}
\note{
    The direct cross-validation of \code{\link{gssanova}} can be more
    effective in general, and more stable for complex models.

    For large sample sizes, the approximate solution of
    \code{\link{gssanova}} can be faster.

    The method \code{\link{project}} is not implemented for
    \code{gssanova0}, nor is the mixed-effect model support through
    \code{\link{mkran}}.

    In \emph{gss} versions earlier than 1.0, \code{gssanova0} was under
    the name \code{gssanova}.
}
\author{Chong Gu, \email{chong@stat.purdue.edu}}
\references{
    Gu, C. (1992), Cross-validating non Gaussian data. \emph{Journal
	of Computational and Graphical Statistics}, \bold{1}, 169-179.
}
\examples{
## Fit a cubic smoothing spline logistic regression model
test <- function(x)
        {.3*(1e6*(x^11*(1-x)^6)+1e4*(x^3*(1-x)^10))-2}
x <- (0:100)/100
p <- 1-1/(1+exp(test(x)))
y <- rbinom(x,3,p)
logit.fit <- gssanova0(cbind(y,3-y)~x,family="binomial")
## The same fit
logit.fit1 <- gssanova0(y/3~x,"binomial",weights=rep(3,101))
## Obtain estimates and standard errors on a grid
est <- predict(logit.fit,data.frame(x=x),se=TRUE)
## Plot the fit and the Bayesian confidence intervals
plot(x,y/3,ylab="p")
lines(x,p,col=1)
lines(x,1-1/(1+exp(est$fit)),col=2)
lines(x,1-1/(1+exp(est$fit+1.96*est$se)),col=3)
lines(x,1-1/(1+exp(est$fit-1.96*est$se)),col=3)
## Clean up
\dontrun{rm(test,x,p,y,logit.fit,logit.fit1,est)
dev.off()}
}
\keyword{models}
\keyword{regression}
\keyword{smooth}