Bootstrap, jackknife, and other objects that inherit from "resamp".

DESCRIPTION:

Classes that inherit from resamp include: bootstrap, bootstrap2, concomitants.bootstrap, controlVariates.bootstrap, influence, jackknife, limits.abc, parametricBootstrap, parametricBootstrapTest, permutationTest, permutationTest2, permutationTestMeans, reweight, smoothedBootstrap.

GENERATION:

These objects are created by functions of the same names (except for "resamp", which is the parent class, for which there is no function.

METHODS:

The "resamp" class of objects has methods for a large number of generic functions, see below.

VALUE:

The following components must be included in a legitimate resamp object.
call
the call that produced the object
observed
the value of the statistic, calculated on the original data. This is saved as a vector with names, say of length p.
replicates
matrix of dimensions B by p, where B is the number of resamples.
estimate
data frame with p rows, summarizing the results. and objects have columns "Mean", "Bias" (a bias estimate), and "SE" (standard error estimate). Other "resamp" objects have other columns.
B
integer, total number of resamples. Note that and other functions may accept an B as vector, e.g. for importance sampling. That information is stored in the call, and only the total number of resamples is stored here.
n
integer, number of observations in the original data.
dim.obs
dimension of the observed statistic

OTHER COMPONENTS:

VALUE:

Additional components are either required or optional in various objects that inherit from "resamp", including:

label
character, used when printing or plotting.
defaultLabel
character, used when plotting if label not present.
group
vector, the unique values of this vector determine groups (for stratified sampling or multiple-sample problems).
subject
vector, the unique values of this vector determine subjects (for sampling by subject).
treatment
vector, the unique values of this vector determine two groups.
seed.start
the value of the seed for initializing the sampler. For random sampling this is typically a legal input to set.seed, i.e. an integer from 0 to 1023 or a value of .Random.seed.
seed.end
object describing that state of sampling at the end of sampling; this may be used as the input seed for adding additional samples. For random sampling this is typically the value of .Random.seed after sampling.
B.missing
integer, number of resamples with missing values (in any of the p components).
indices
matrix with dimensions n by B, vector of indices for resampling.
compressedIndices
object of class , a version of the indices.
weights
vector of length B of weights that result from e.g. importance sampling. Summaries of results should use these weights.
L
matrix with dimensions n by p, used for creating linear approximations to a statistic.
Lstar
matrix with dimensions B by p, linear approximations for the resamples.
n.groups
number of groups of observations left out by .
original
list containing original parts of a "resamp" object, before they are replaced by a procedure such as that improves accuracy.
statistic
original statistic, as modified to support weights.
actual.calls
a list whose components are call objects. Some resampling functions such as manipulate the call component of an object; the original call(s) may be added as a component of this list.
parent.frame
generally a frame number. This is used as the default for frame.eval for many functions. This is used when the resamp object was created within another function.
bootstrap.objects
a list of two bootstrap objects, created by and each corresponding to one value of the treatment.

Other components used by specific functions:

rsampler
the rsampler argument to .
args.rsampler
the args.rsampler argument to .
dsampler
the dsampler argument to .
samples
for , actual samples used.

SEE ALSO:

The following functions create "resamp" objects: , , , , , , , , , , and .

These functions modify "resamp" objects, to improve accuracy or perform diagnostics: , , , and .

These functions use "resamp" objects for inferences or diagnostics: , , , , , , , and .

The following generic functions have methods for "resamp" objects, and in some cases for child classes: , , (and Maxs, Mins, Ranges, Sums, Vars), , , , , , , , , (and Maxs, Mins, Ranges, Sums, Vars), , and .

Some other functions that operate on "resamp" objects, or parts of them: , , , and .

For an annotated list of functions in the package, including other high-level resampling functions, see: .