monteBootBB {ssExtra}R Documentation

Calculate Bootstrap and Jackkife Statistics on an Individual Monte Carlo Sample

Description

This routine will perform bootstrap resampling as well as the jackknife on a single Monte Carlo sample replicate drawn from the sampling surfaces of "ssBigBAF" objects. This routine is called from within the nested sample size–within–Monte Carlo loops in the "monteBigBAF" constructor. However, it certainly can be called in other situations where the correct information is available. Among other things, it calculates the ‘bca’ confidence intervals on the replicate sample.

Usage

monteBootBB(samples,
            B = 100,
            alpha = 0.05,
            ...)

Arguments

samples

This must be a samples data frame calculated in a single Monte Carlo draw as returned from a call to monteStatsBB for a given sample size, n.

B

The number of bootstrap samples.

alpha

The two-tailed alpha level, used to select the correct bootstrap ‘bca’ intervals for comparison.

...

Gobbled for now.

Details

This routine uses the bcajack routine from bcaboot for construction of both bootstrap and jackknife confidence intervals for a Monte Carlo sample from an "ssBigBAF" object. The calculations in that package are all done in native R code, and so can be slow when applied to many Monte Carlo replicates in the context of big BAF sampling. Please see the bcaboot package for more details, and also the reference below.

Value

A list invisibly as described in the return from bcajack...

call

The call to bcajack as returned from match.call

lims

“first column shows the estimated bca confidence limits...”

stats

“top line of stats shows 5 estimates...”

B.mean

“bootstrap sample size B, and the mean...”

ustats

“The bias-corrected estimator...”

seed

“The random number state for reproducibility”

including the following as part of the list...

jackmean

The jackknife estimate of the mean.

Author(s)

Jeffrey H. Gove

References

Efron, B. and B. Narasimhan. 2018. The automatic construction of bootstrap confidence intervals. Technical Report No. 2018-07, Standford University Department of Statistics, PDF version: here.

See Also

Please also see the testBoot routine for some more information on a comparison of the results from the bcaboot package to those in the boot package.

Examples

#
# If 'stats' is a return object from monteStatsBB, then
# the following will draw 200 bootstrap samples and also
# calculate the jackknife results...
#
## Not run: 
mcb = monteBootBB(stats$samples, 200)

## End(Not run) 

[Package ssExtra version 0.1-2 Index]