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
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
|
\name{summary.lmrob}
\title{Summary Method for "lmrob" Objects}
%
\alias{summary.lmrob}
\alias{hatvalues.lmrob}
\alias{.lmrob.hat}
\alias{vcov.lmrob}
\alias{print.summary.lmrob}
\alias{model.matrix.lmrob}
%
\description{
Summary method for \R object of class \code{"lmrob"} and
\code{\link{print}} method for the summary object.
Further, methods \code{\link{fitted}()}, \code{\link{residuals}()}
work (via the default methods), and
\code{\link{predict}()} (see \code{\link{predict.lmrob}},
\code{\link{vcov}()}, \code{\link{weights}()} (see
\code{\link{weights.lmrob}}), \code{\link{model.matrix}()},
\code{\link{confint}()}, \code{\link{dummy.coef}()},
\code{\link{hatvalues}()}, etc.,
have explicitly defined \code{lmrob} methods. \code{.lmrob.hat()} is
the lower level \dQuote{work horse} of the \code{hatvalues()} method.
}
\usage{% all source in ../R/lmrob.R <<<
\method{summary}{lmrob}(object, correlation = FALSE,
symbolic.cor = FALSE, \dots)
\method{print}{summary.lmrob}(x, digits = max(3, getOption("digits") - 3),
symbolic.cor= x$symbolic.cor,
signif.stars = getOption("show.signif.stars"),
showAlgo = TRUE, \dots)
\method{vcov}{lmrob}(object, cov = object$control$cov, complete = TRUE, \dots)
\method{model.matrix}{lmrob}(object, \dots)
% not yet
% .lmrob.hat(x, w = rep(1, NROW(x)), wqr = qr(sqrt(w) * x))
}
\arguments{
\item{object}{an \R object of class \code{lmrob}, typically created by
\code{\link{lmrob}}.}
\item{correlation}{logical variable indicating whether
to compute the correlation matrix of the estimated coefficients.}
\item{symbolic.cor}{logical indicating whether
to use symbols to display the above correlation matrix.}
\item{x}{an \R object of class \code{summary.lmrob}, typically
resulting from \code{summary(\link{lmrob}(..),..)}.}
\item{digits}{number of digits for printing, see \code{digits} in
\code{\link{options}}.}
\item{signif.stars}{logical variable indicating
whether to use stars to display different levels of
significance in the individual t-tests.}
\item{showAlgo}{optional \code{\link{logical}} indicating if the
algorithmic parameters (as mostly inside the \code{control} part)
should be shown.}
\item{cov}{covariance estimation function to use, a
\code{\link{function}} or \link{character} string naming the
function; \pkg{robustbase} currently provides \code{".vcov.w"} and
\code{".vcov.avar1"}, see \emph{Details} of \code{\link{lmrob}}.
Particularly useful when \code{object} is the result of
\code{lmrob(.., cov = "none")}, where \preformatted{ object$cov <- vcov(object, cov = ".vcov.w")}
allows to \emph{update} the fitted object.}
\item{complete}{(mainly for \R \code{>= 3.5.0}:)% ~/R/D/r-devel/R/src/library/stats/man/vcov.Rd
\code{\link{logical}} indicating if the
full variance-covariance matrix should be returned also in case of
an over-determined system where some coefficients are undefined and
\code{\link{coef}(.)} contains \code{NA}s correspondingly. When
\code{complete = TRUE}, \code{vcov()} is compatible with
\code{coef()} also in this singular case.}
\item{\dots}{potentially more arguments passed to methods.}
}
\value{
\code{summary(object)} returns an object of S3 class
\code{"summary.lmrob"}, basically a \code{\link{list}} with components
"call", "terms", "residuals", "scale", "rweights", "converged",
"iter", "control" all copied from \code{object}, and further
components, partly for compatibility with \code{\link{summary.lm}},
\item{coefficients}{a \code{\link{matrix}} with columns \code{"Estimate"},
\code{"Std. Error"}, \code{"t value"}, and \code{"PR(>|t|)"}, where
"Estimate" is identical to \code{\link{coef}(object)}. Note that
\code{\link{coef}(<summary.obj>)} is slightly preferred to access
this matrix.}
\item{df}{degrees of freedom, in an \code{\link{lm}} compatible way.}
\item{sigma}{identical to \code{\link{sigma}(object)}.}
\item{aliased}{..}%FIXME
\item{cov}{derived from \code{object$cov}.}% FIXME: say more
\item{r.squared}{robust \dQuote{R squared} or \eqn{R^2}, a coefficient
of determination: This is the consistency corrected robust
coefficient of determination by Renaud and Victoria-Feser (2010).}
\item{adj.r.squared}{an adjusted R squared, see \code{r.squared}.}
}
\references{
Renaud, O. and Victoria-Feser, M.-P. (2010).
A robust coefficient of determination for regression,
\emph{Journal of Statistical Planning and Inference} \bold{140}, 1852-1862.
}
\seealso{\code{\link{lmrob}}, \code{\link{predict.lmrob}},
\code{\link{weights.lmrob}}, \code{\link{summary.lm}},
\code{\link{print}}, \code{\link{summary}}.
}
\examples{
mod1 <- lmrob(stack.loss ~ ., data = stackloss)
sa <- summary(mod1) # calls summary.lmrob(....)
sa # dispatches to call print.summary.lmrob(....)
## correlation between estimated coefficients:
cov2cor(vcov(mod1))
cbind(fit = fitted(mod1), resid = residuals(mod1),
wgts= weights(mod1, type="robustness"),
predict(mod1, interval="prediction"))
data(heart)
sm2 <- summary( m2 <- lmrob(clength ~ ., data = heart) )
sm2
}
\keyword{robust}
\keyword{regression}
|
