MINI MINI MANI MO

%
% Copyright (c) 2012, 2015, Oracle and/or its affiliates. All rights reserved.
%
\name{ore.lm}
\alias{ore.lm}
\alias{ore.stepwise}
\alias{summary.ore.lm}
\alias{logLik.ore.lm}
\alias{hatvalues.ore.lm}
\alias{vcov.ore.lm}
\alias{print.summary.ore.lm}
\alias{print.ore.stepwise}
\alias{predict.ore.lm}
\alias{plot.ore.lm}
\alias{drop1.ore.lm}
\alias{add1.ore.lm}
\concept{regression}
\title{Oracle R Enterprise Linear and Stepwise Linear Regression Models}
\description{
  Model functions for fitting linear regression and stepwise linear
  regression models on \code{ore.frame} data.
}
\usage{
### Fitting functions
ore.lm(formula, data, weights, contrasts = NULL, xlev = NULL,
       ...)
ore.stepwise(formula, data, scope,
             direction = c("both", "backward", "forward", "alternate", "none"),
             add.p = 0.50, drop.p = 0.10, nbest = 1, steps = 1000,
             contrasts = NULL, xlev = NULL, ...)

### Specific methods for ore.lm objects
\S3method{summary}{ore.lm}(object, correlation = FALSE, symbolic.cor = FALSE, ...)

\S3method{logLik}{ore.lm}(object, REML = FALSE, ...)

\S3method{hatvalues}{ore.lm}(model, ...)

\S3method{vcov}{ore.lm}(object, ...)

\S3method{predict}{ore.lm}(object, newdata, se.fit = FALSE, scale = NULL, df = Inf, 
        interval = c("none", "confidence", "prediction"),
        level = 0.95, na.action = na.pass, pred.var = NULL,
        weights = NULL, supplemental.cols = NULL, ...)

\S3method{add1}{ore.lm}(object, scope, scale = 0, test = c("none", "Chisq", "F"),
     x = NULL, k = 2, ...)

\S3method{drop1}{ore.lm}(object, scope, scale = 0, all.cols = TRUE,
      test = c("none", "Chisq", "F"), k = 2, ...)

### Inherited methods for ore.lm objects
#anova(object, ...)
#coef(object, ...)
#coefficients(object, ...)
#confint(object, parm, level = 0.95, ...)
#deviance(object, ...)
#effects(object, ...)
#extractAIC(fit, scale, k = 2, ...)
#fitted(object, ...)
#fitted.values(object, ...)
#formula(x, ...)
#model.frame(formula, ...)
#nobs(object, ...)
#resid(object, ...)
#residuals(object, ...)
#weights(object, ...)
}
\arguments{
  \item{formula}{A \code{\link[stats]{formula}} object representing the
    model (\code{ore.lm}) or initial model (\code{ore.stepwise}) to be
    fit. This formula must be of the form \code{response ~ terms}, where
    the \code{response} represents a single variable and the
    \code{terms} are derived from \code{ore.number} and
    \code{ore.factor} columns from the \code{data} argument.}
  \item{data}{An \code{ore.frame} object specifying the data for the
    model. White spaces are not supported in any of the column names
    used in the \code{formula} argument.}
  \item{weights}{In function \code{ore.lm} an optional \code{ore.number}
    object specifying the analytic weights in the model.
    In function \code{predict.ore.lm} when argument \code{interval} is
    \code{"prediction"} and argument \code{pred.val} is \code{NULL} and
    \code{object$weights} is not \code{NULL}, the variance weights
    for the predictions as either an
    \code{\link[OREbase:ore.numeric-class]{ore.numeric}} object or a
      one-sided model \code{\link[stats]{formula}} referring to data
      within argument \code{newdata}.}
  \item{ supplemental.cols }{ Additional columns to include in the prediction
  result from the \code{newdata} data set. }
  \item{contrasts}{An optional named \code{\link[base]{list}} to be
    supplied to the \code{contrasts.arg} argument of
    \code{\link[stats]{model.matrix}}.}
  \item{xlev}{An optional named \code{\link[base]{list}} of
    \code{\link[base]{character}} vectors specifying the
    \code{\link[base]{levels}} for each
    \code{\link[OREbase:ore.factor-class]{ore.factor}} variable.}
  \item{scope}{The range of models to examine within the stepwise
    procedure; either a single \code{\link[stats]{formula}} object
    representing the upper model or a list object containing
    \code{"lower"} and \code{"upper"} \code{\link[stats]{formula}}
    object elements.}
  \item{direction}{The stepwise search mode; one of \code{"both"} (first
    try to add a term using the \code{add.p} argument value and then try
    repeatedly to drop terms using the \code{drop.p} argument value), 
    \code{"backward"}, \code{"forward"}, \code{"alternate"} (similar to
    \code{"both"} but only one drop is attempted per add attempt) or
    \code{"none"} with a default of \code{"both"}.}
  \item{add.p}{The F-test p-value threshold for adding a term to the
    model.}
  \item{drop.p}{The F-test p-value threshold for removing a term from
    the model.}
  \item{nbest}{The number of best models, according to the selection
    criteria, to report at each step.}
  \item{steps}{The maximum number of steps to make.}
  \item{object, model, newdata}{\code{ore.lm} object.}
  \item{correlation, symbolic.cor}{Argument not implemented.}
  \item{REML}{Argument not implemented.}
  \item{se.fit}{A logical value indicating whether to return the
    standard errors for the predictions.}
  \item{scale}{The scale parameter for standard error of the predictions.}
  \item{df}{The degrees of freedom for the predictions when argument
    \code{scale} is not \code{NULL}.}
  \item{interval}{The type of interval to return, either \code{"none"},
    \code{"confidence"}, or \code{"prediction"}.}
  \item{level}{The level for argument \code{interval}.}
  \item{na.action}{The manner in which \code{NA} values are handled,
    either \code{na.omit} or \code{na.pass}.}
  \item{pred.var}{When argument \code{interval} is \code{"prediction"},
    the variance for a single observation.}
  \item{test, x, k, all.cols}{See function \code{\link[stats]{add1}}.}
  \item{\dots}{Additional arguments.}
}
\details{
  The \code{ore.lm} and \code{ore.stepwise} functions perform least
  squares regression and stepwise least squares regression
  respectively on data represented in \code{ore.frame} objects.

  A model fit is generated using embedded R map/reduce operations where
  the map operation creates either QR decompositions or matrix
  cross-products depending on the number of coefficients being
  estimated. The underlying model matrices will be created using either
  \code{model.matrix} or \code{sparse.model.matrix} depending on its
  sparsity. Once the coefficients for the model have been estimated
  another pass of the data is made to estimate the model-level statistics.

  When forward, backward, or stepwise selection is performed, the XtX
  and Xty matrices are subsetted to generate the F-test p-values based
  upon coefficient estimates that were generated using a Choleski
  decomposition of the XtX subset matrix.

  If there are collinear terms in the model, functions \code{ore.lm} and
  \code{ore.stepwise} will not estimate the coefficient values for
  a collinear set of terms. In the case of \code{ore.stepwise},
  this collinear set of terms will be excluded throughout the
  procedure.

  The \code{\link[OREbase:ore.options]{"ore.parallel"}} global option is
  used by \code{ore.lm} to determine the preferred degree of
  parallelism to use within the Oracle R Enterprise server.
}
\value{
  For \code{ore.lm}, returns an \code{ore.lm} object.

  For \code{summary.ore.lm}, returns a \code{summary.ore.lm} object.

  For \code{ore.stepwise}, returns an \code{ore.stepwise} object.

  Note: the training data referenced by argument \code{data} is needed to
  produce meta information about the \code{ore.lm} object.
}
\references{
  \href{http://www.oracle.com/technetwork/database/database-technologies/r/r-enterprise/documentation/index.html}{Oracle R Enterprise}
}
\author{
  Oracle \email{oracle-r-enterprise@oracle.com}
}
\seealso{
  \link[OREstats]{model.matrix,formula-method} (\pkg{OREstats} package),
  \code{\link{ore.glm}},
  \code{\link[stats]{lm}},
  \code{\link[stats]{step}},
  \code{\link[OREbase:ore.options]{ore.parallel}}
}
\examples{
  # Create database table
  library(OREstats)
  LONGLEY <- ore.push(longley)

  # Fit full model
  oreFit1 <- ore.lm(Employed ~ ., data = LONGLEY)
  summary(oreFit1)

  # Two stepwise alternatives
  oreStep1 <-
    ore.stepwise(Employed ~ .^2, data = LONGLEY, add.p = 0.1, drop.p = 0.1)
  oreStep2 <-
    step(ore.lm(Employed ~ 1, data = LONGLEY),
         scope = terms(Employed ~ .^2, data = LONGLEY))
}
\keyword{regression}