These functions are used by the Bootstrap dialog.
menuBootstrap calls
tabSummary.bootstrap
,
tabPlot.bootstrap and
tabJab.bootstrap if
summary, plotting and jackknife-after-bootstrap results are requested.
data to be bootstrapped. May be a vector, matrix, or data frame.
statistic
statistic to be bootstrapped. It may be a function which accepts data
as the first argument and returns a vector or matrix, or a call
referring to the data which evaluates to a vector or matrix.
If it is an expression and the data argument is the name
of an object, then the object should also be referred to by name in
the expression. If the data are constructed within the call to
bootstrap, then data should be referred to as data in the expression.
(See examples for
bootstrap.)
boot.obj
an object of class
bootstrap.
OPTIONAL ARGUMENTS:
B
number of bootstrap resamples to be drawn. We recommend at least 250
to estimate standard errors and 1000 to estimate percentiles.
group
allows stratified sampling and bootstrapping multi-sample problems.
The unique values of this vector determine groups. For each resample,
a bootstrap sample is drawn separately for each group, and the
observations are combined to give the full resample. The statistic is
calculated for the resample as a whole.
seed
seed for generating resampling indices. May be a legal random number
seed or an integer between 0 and 1000 which will be passed to
set.seed.
block.size
control variable specifying the number of resamples to calculate at
once.
bootstrap uses an
lapply() within a
for() loop. For small
sample sizes, a single
lapply() is reasonable, while for large sample
sizes, a series of separate
lapply()s is more efficient. Using
sampler=samp.boot.bal gives balancing done separately within each
group of resamples.
trace
logical flag indicating whether the algorithm should print a message
indicating which set of replicates is currently being drawn.
assign.frame1
logical flag indicating whether the resampled data should be assigned
to frame 1 before evaluating the statistic. This may be necessary if
the statistic is reevaluating the call of a model object. If all
bootstrap estimates are identical, try setting
assign.frame1=T. Note
that this will slow down the algorithm.
save.indices
logical flag indicating whether to save the matrix of resampling indices.
probs
probability levels at which to calculate percentiles of confidence intervals.
print.short.p
logical flag indicating whether to print basic summary statistics.
print.emp.p
logical flag indicating whether to print empirical percentiles.
print.bca.p
logical flag indicating whether to print BCa percentiles.
print.cor.p
logical flag indicating whether to print correlation matrix of statistic
if statistic is a vector.
plot.p
logical flag indicating whether to plot the replicates with
plot.resamp.
plotQQ.p
logical flag indicating whether to plot a QQ-plot of the replicates
with
qqnorm.resamp.
functional.jab
functional of the bootstrap distribution for which standard error is
to be estimated when using jackknife-after-bootstrap. May be a
character string specifying a column of
boot.obj$stats (i.e.
"Bias",
"Mean", or
"SE") or a function expecting a vector as its first
argument and returning a scalar, e.g.
max.
print.jab.p
logical flag indicating whether to print jackknife-after-bootstrap summaries.
plot.jab.p
logical flag indicating whether to plot a jackknife-after-bootstrap influence
plot.
save.name.jab
name under which to save the
jack.after.bootstrap object produced by
the jackknife-after-bootstrap analysis.
...
other arguments to pass to the
bootstrap function.
VALUE:
menuBootstrap returns an object of class
bootstrap. See the
bootstrap
help files for details.
tabSummary.bootstrap and
tabPlot.bootstrap return
the
boot.obj.
tabJab.bootstrap returns the
jack.after.bootstrap object.
SIDE EFFECTS:
If
save.name.jab is specified, the jack-after-bootstrap is assigned to the
working database under the specified name.