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
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
|
% File nlme/man/plot.gls.Rd
% Part of the nlme package for R
% Distributed under GPL 2 or later: see nlme/LICENCE.note
\name{plot.gls}
\title{Plot a gls Object}
\usage{
\method{plot}{gls}(x, form, abline, id, idLabels, idResType, grid, \dots)
}
\alias{plot.gls}
\arguments{
\item{x}{an object inheriting from class \code{"\link{gls}"}, representing
a generalized least squares fitted linear model.}
\item{form}{an optional formula specifying the desired type of
plot. Any variable present in the original data frame used to obtain
\code{x} can be referenced. In addition, \code{x} itself
can be referenced in the formula using the symbol
\code{"."}. Conditional expressions on the right of a \code{|}
operator can be used to define separate panels in a Trellis
display. Default is \code{resid(., type = "p") ~ fitted(.) },
corresponding to a plot of the standardized residuals versus fitted
values, both evaluated at the innermost level of nesting.}
\item{abline}{an optional numeric value, or numeric vector of length
two. If given as a single value, a horizontal line will be added to the
plot at that coordinate; else, if given as a vector, its values are
used as the intercept and slope for a line added to the plot. If
missing, no lines are added to the plot.}
\item{id}{an optional numeric value, or one-sided formula. If given as
a value, it is used as a significance level for a two-sided outlier
test for the standardized residuals. Observations with
absolute standardized residuals greater than the \eqn{1 - value/2}
quantile of the standard normal distribution are identified in the
plot using \code{idLabels}. If given as a one-sided formula, its
right hand side must evaluate to a logical, integer, or character
vector which is used to identify observations in the plot. If
missing, no observations are identified.}
\item{idLabels}{an optional vector, or one-sided formula. If given as a
vector, it is converted to character mode and used to label the
observations identified according to \code{id}. If given as a
one-sided formula, its right hand side must evaluate to a vector
which is converted to character mode and used to label the identified
observations. Default is the innermost grouping factor.}
\item{idResType}{an optional character string specifying the type of
residuals to be used in identifying outliers, when \code{id} is a
numeric value. If \code{"pearson"}, the standardized residuals (raw
residuals divided by the corresponding standard errors) are used;
else, if \code{"normalized"}, the normalized residuals (standardized
residuals pre-multiplied by the inverse square-root factor of the
estimated error correlation matrix) are used. Partial matching of
arguments is used, so only the first character needs to be
provided. Defaults to \code{"pearson"}.}
\item{grid}{an optional logical value indicating whether a grid should
be added to plot. Default depends on the type of Trellis plot used:
if \code{xyplot} defaults to \code{TRUE}, else defaults to
\code{FALSE}.}
\item{\dots}{optional arguments passed to the Trellis plot function.}
}
\description{
Diagnostic plots for the linear model fit are obtained. The
\code{form} argument gives considerable flexibility in the type of
plot specification. A conditioning expression (on the right side of a
\code{|} operator) always implies that different panels are used for
each level of the conditioning factor, according to a Trellis
display. If \code{form} is a one-sided formula, histograms of the
variable on the right hand side of the formula, before a \code{|}
operator, are displayed (the Trellis function \code{histogram} is
used). If \code{form} is two-sided and both its left and
right hand side variables are numeric, scatter plots are displayed
(the Trellis function \code{xyplot} is used). Finally, if \code{form}
is two-sided and its left had side variable is a factor, box-plots of
the right hand side variable by the levels of the left hand side
variable are displayed (the Trellis function \code{bwplot} is used).
}
\value{
a diagnostic Trellis plot.
}
\author{José Pinheiro and Douglas Bates \email{bates@stat.wisc.edu}}
\seealso{\code{\link{gls}}, \code{\link[lattice]{xyplot}},
\code{\link[lattice]{bwplot}}, \code{\link[lattice]{histogram}}
}
\examples{
fm1 <- gls(follicles ~ sin(2*pi*Time) + cos(2*pi*Time), Ovary,
correlation = corAR1(form = ~ 1 | Mare))
# standardized residuals versus fitted values by Mare
plot(fm1, resid(., type = "p") ~ fitted(.) | Mare, abline = 0)
# box-plots of residuals by Mare
plot(fm1, Mare ~ resid(.))
# observed versus fitted values by Mare
plot(fm1, follicles ~ fitted(.) | Mare, abline = c(0,1))
}
\keyword{models}
|
