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
|
\name{BYlogreg}
\alias{BYlogreg}
\title{Bianco-Yohai Estimator for Robust Logistic Regression}
\encoding{utf8}
\description{
Computation of the estimator of Bianco and Yohai (1996) in logistic regression.
Now provides both the \emph{weighted} and regular (unweighted) BY-estimator.
By default, an intercept term is included and p parameters are estimated.
For more details, see the reference.
Note: This function is for \dQuote{back-compatibility} with the
\code{BYlogreg()} code web-published at KU Leuven, Belgium,
% moved to ? -- but NA (404) on 2017-09-26: % --> ../R/BYlogreg.R
% \url{http://feb.kuleuven.be/public/NDBAE06/programs/roblog/};
% now, findable at
% http://feb.kuleuven.be/public/u0017833/software_donotuse/logreg/BYlogreg.txt
% but rather use Wiley's book resources
and also available as file \file{FunctionsRob/BYlogreg.ssc} from
\url{https://www.wiley.com/legacy/wileychi/robust_statistics/robust.html}.
However instead of using this function, the recommended interface is
\code{\link{glmrob}(*, method = "BY")} or \code{... method = "WBY" ..},
see \code{\link{glmrob}}.
}
\usage{
BYlogreg(x0, y, initwml = TRUE, addIntercept = TRUE,
const = 0.5, kmax = 1000, maxhalf = 10, sigma.min = 1e-4,
trace.lev = 0)
}
\arguments{
\item{x0}{a numeric \eqn{n \times (p-1)}{n * (p-1)} matrix containing
the explanatory variables.}
\item{y}{numeric \eqn{n}-vector of binomial (0 - 1) responses.}
\item{initwml}{logical for selecting one of the two possible methods
for computing the initial value of the optimization process.
If \code{initwml} is true (default), a weighted ML estimator is
computed with weights derived from the MCD estimator
computed on the explanatory variables.
If \code{initwml} is false, a classical ML fit is perfomed. When
the explanatory variables contain binary observations, it is
recommended to set initwml to FALSE or to modify the code of the
algorithm to compute the weights only on the continuous variables.
}
\item{addIntercept}{logical indicating that a column of \code{1} must be
added the \eqn{x} matrix.}
\item{const}{tuning constant used in the computation of the estimator
(default=0.5).}
\item{kmax}{maximum number of iterations before convergence (default=1000).}
\item{maxhalf}{max number of step-halving (default=10).}
\item{sigma.min}{smallest value of the scale parameter before
implosion (and hence non-convergence) is assumed.}
\item{trace.lev}{logical (or integer) indicating if intermediate results
should be printed; defaults to \code{0} (the same as \code{FALSE}).}
}
%% \details{
%% If necessary, more details than the description above
%% }
\value{
a list with components
\item{convergence}{logical indicating if convergence was achieved}
\item{objective}{the value of the objective function at the minimum}
\item{coefficients}{vector of parameter estimates}
\item{vcov}{variance-covariance matrix of the coefficients (if convergence is TRUE).}
\item{sterror}{standard errors, i.e., simply \code{sqrt(diag(.$vcov))}, if convergence.}
}
\references{
Croux, C., and Haesbroeck, G. (2003)
Implementing the Bianco and Yohai estimator for Logistic Regression,
\emph{Computational Statistics and Data Analysis} \bold{44}, 273--295.
Ana M. Bianco and Víctor J. Yohai (1996)
Robust estimation in the logistic regression model.
In Helmut Rieder, \emph{Robust Statistics, Data Analysis, and
Computer Intensive Methods}, Lecture Notes in Statistics \bold{109},
pages 17--34.
}
\author{
Originally, Christophe Croux and Gentiane Haesbroeck, with
thanks to Kristel Joossens and Valentin Todorov for improvements.
Speedup, tweaks, more \dQuote{control} arguments: Martin Maechler.
}
\seealso{
The more typical way to compute BY-estimates (via
\code{\link{formula}} and methods):
\code{\link{glmrob}(*, method = "WBY")} and \code{.. method = "BY"}.
}
\examples{
set.seed(17)
x0 <- matrix(rnorm(100,1))
y <- rbinom(100, size=1, prob= 0.5) # ~= as.numeric(runif(100) > 0.5)
BY <- BYlogreg(x0,y)
BY <- BYlogreg(x0,y, trace.lev=TRUE)
## The "Vaso Constriction" aka "skin" data:
data(vaso)
vX <- model.matrix( ~ log(Volume) + log(Rate), data=vaso)
vY <- vaso[,"Y"]
head(cbind(vX, vY))# 'X' does include the intercept
vWBY <- BYlogreg(x0 = vX, y = vY, addIntercept=FALSE) # as 'vX' has it already
v.BY <- BYlogreg(x0 = vX, y = vY, addIntercept=FALSE, initwml=FALSE)
## they are relatively close, well used to be closer than now,
## with the (2023-05, VT) change of covMcd() scale-correction
stopifnot( all.equal(vWBY, v.BY, tolerance = 0.008) ) # was ~ 1e-4 till 2023-05
}
\keyword{robust}
\keyword{regression}
\keyword{nonlinear}
|
