1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
|
% $Id: comparePred.Rd,v 1.5.2.1 2003/04/17 22:28:45 bates Exp $
\name{comparePred}
\title{Compare Predictions}
\usage{
comparePred(object1, object2, primary, minimum, maximum,
length.out, level, \dots)
}
\alias{comparePred}
\alias{comparePred.gls}
\alias{comparePred.lme}
\alias{comparePred.lmList}
\arguments{
\item{object1,object2}{fitted model objects, from which predictions can
be extracted using the \code{predict} method.}
\item{primary}{an optional one-sided formula specifying the primary
covariate to be used to generate the augmented predictions. By
default, if a covariate can be extracted from the data used to generate
the objects (using \code{getCovariate}), it will be used as
\code{primary}.}
\item{minimum}{an optional lower limit for the primary
covariate. Defaults to \code{min(primary)}, after \code{primary} is
evaluated in the \code{data} used in fitting \code{object1}.}
\item{maximum}{an optional upper limit for the primary
covariate. Defaults to \code{max(primary)}, after \code{primary} is
evaluated in the \code{data} used in fitting \code{object1}.}
\item{length.out}{an optional integer with the number of primary
covariate values at which to evaluate the predictions. Defaults to
51.}
\item{level}{an optional integer specifying the desired
prediction level. Levels increase from outermost to innermost
grouping, with level 0 representing the population (fixed effects)
predictions. Only one level can be specified. Defaults to the
innermost level.}
\item{\dots}{some methods for the generic may require additional
arguments.}
}
\description{
Predicted values are obtained at the specified values of
\code{primary} for each object. If either \code{object1} or
\code{object2} have a grouping structure
(i.e. \code{getGroups(object)} is not \code{NULL}), predicted values
are obtained for each group. When both objects determine groups, the
group levels must be the same. If other covariates besides
\code{primary} are used in the prediction model, their group-wise averages
(numeric covariates) or most frequent values (categorical covariates)
are used to obtain the predicted values. The original observations are
also included in the returned object.
}
\value{
a data frame with four columns representing, respectively, the values
of the primary covariate, the groups (if \code{object} does not have a
grouping structure, all elements will be \code{1}), the predicted or
observed values, and the type of value in the third column: the
objects' names are used to classify the predicted values and
\code{original} is used for the observed values. The returned object
inherits from classes \code{comparePred} and \code{augPred}.
}
\author{Jose Pinheiro \email{Jose.Pinheiro@pharma.novartis.com} and Douglas Bates \email{bates@stat.wisc.edu}}
\note{
This function is generic; method functions can be written to handle
specific classes of objects. Classes which already have methods for
this function include: \code{gls}, \code{lme}, and \code{lmList}.
}
\seealso{\code{\link{augPred}}, \code{\link{getGroups}}}
\examples{
fm1 <- lme(distance ~ age * Sex, data = Orthodont, random = ~ age)
fm2 <- update(fm1, distance ~ age)
comparePred(fm1, fm2, length.out = 2)
}
\keyword{models}
|
