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
|
\name{boot.array}
\alias{boot.array}
\title{
Bootstrap Resampling Arrays
}
\description{
This function takes a bootstrap object calculated by one of the
functions \code{boot}, \code{censboot}, or \code{tilt.boot} and
returns the frequency (or index) array for the bootstrap
resamples.
}
\usage{
boot.array(boot.out, indices=)
}
\arguments{
\item{boot.out}{
An object of class \code{"boot"} returned by one of the generation
functions for such an object.
}
\item{indices}{
A logical argument which specifies whether to return the frequency
array or the raw index array. The default is \code{indices=FALSE}
unless \code{boot.out} was created by \code{tsboot} in which case the
default is \code{indices=TRUE}.
}}
\value{
A matrix with \code{boot.out$R} rows and \code{n} columns where
\code{n} is the number of observations in \code{boot.out$data}. If
\code{indices} is \code{FALSE} then this will give the frequency of
each of the original observations in each bootstrap resample. If
\code{indices} is \code{TRUE} it will give the indices of the
bootstrap resamples in the order in which they would have been passed
to the statistic.
}
\section{Side Effects}{
This function temporarily resets \code{.Random.seed} to the value in
\code{boot.out$seed} and then returns it to its original value at the
end of the function.
}
\details{
The process by which the original index array was generated is
repeated with the same value of \code{.Random.seed}. If the frequency
array is required then \code{freq.array} is called to convert the
index array to a frequency array.
A resampling array can only be returned when such a concept makes
sense. In particular it cannot be found for any parametric or
model-based resampling schemes. Hence for objects generated by
\code{censboot} the only resampling scheme for which such an array can
be found is ordinary case resampling. Similarly if \code{boot.out$sim}
is \code{"parametric"} in the case of \code{boot} or \code{"model"} in
the case of \code{tsboot} the array cannot be found. Note also that
for post-blackened bootstraps from \code{tsboot} the indices found
will relate to those prior to any post-blackening and so will not be
useful.
Frequency arrays are used in many post-bootstrap calculations such as
the jackknife-after-bootstrap and finding importance sampling weights.
They are also used to find empirical influence values through the
regression method.
}
\seealso{
\code{\link{boot}}, \code{\link{censboot}}, \code{\link{freq.array}},
\code{\link{tilt.boot}}, \code{\link{tsboot}}
}
\examples{
# A frequency array for a nonparametric bootstrap
city.boot <- boot(city, corr, R=40, stype="w")
boot.array(city.boot)
perm.cor <- function(d,i)
cor(d$x,d$u[i])
city.perm <- boot(city, perm.cor, R=40, sim="permutation")
boot.array(city.perm, indices=TRUE)
}
\keyword{nonparametric}
% Converted by Sd2Rd version 1.15.
|
