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
|
\name{weights.lmrob}
\title{Extract Robustness and Model Weights}
\alias{weights.lmrob}
\alias{weights.glmrob}
\description{
\code{weights()} extracts robustness weights or fitting
(or prior) weights from a \code{lmrob} or \code{glmrob} object.
}
\usage{
\method{weights}{lmrob}(object, type = c("prior", "robustness", "working"), ...)
\method{weights}{glmrob}(object, type = c("prior", "robustness", "working"), ...)
}
\arguments{
\item{object}{
an object of class \code{"lmrob"} or \code{"glmrob"}, typically the
result of a call to \code{\link{lmrob}}, or \code{\link{glmrob}},
respectively.}
\item{type}{the type of weights to be returned. Either
\code{"prior"} (default), \code{"robustness"}, or \code{"working"};
working weights for \code{\link{lmrob}} objects are the product of the
prior (often all \eqn{ = 1}) and the robustness weights.
For \code{\link{glmrob}} objects, they currently are the same as the
\code{"robustness"} ones %% not using prior weights as long as not properly implemented there
}
\item{\dots}{not used currently.}
}
\details{
The \dQuote{prior weights} correspond to the weights specified using
the \dQuote{weights} argument when calling \code{lmrob}. The
\dQuote{robustness weights} are the weights assigned by the
M-estimator of regression, \eqn{\psi(r_i/S) / (r_i/S)}. The robust
coefficient estimate then numericarlly corresponds to a weighted least
squares fit using the product of both types of weights as weights;
these are also called \dQuote{working weights}, and are available as
\code{weights(<lmrob>, type = "working")}.
}
\value{
Weights extracted from the object \code{object}.
}
\author{Manuel Koller and Martin Maechler.}
\seealso{
\code{\link{lmrob}}, \code{\link{glmrob}} and \code{\link{weights}}
}
|
