bootstats: Calculate mean and median with bootstrapped confidence...

In AkselA/R-confintplot:

Description

This function returns the bootstrapped mean and median, and the respective confidence intervals, of the supplied numeric vector.

Usage

bootstats(x, p.level = 0.95, nrep = NULL, exact.thrs = 8, smooth = TRUE,
  return_resamples = FALSE)

Arguments

x	A vector of numerical values.
p.level	A number giving the level of confidence for the intervals
nrep	The number of repetitions, or resamples, to perform. Defaults to max(500, 2*(10^5)/length(x)), except if length(x) <= exact.thrs.
exact.thrs	Upper level of length(x) at which all possible subsamples are used

Examples

bootstats(rnorm(1000), p.level=0.9, smooth=FALSE)

bootstats(c(1, 3, 4, 4, 6, 7, 9), exact.thrs=7, smooth=FALSE)

bootstats(c(1, 3, 4, 4, 6, 7, 8), exact.thrs=1, smooth=FALSE)

## Simple smooth function based on jitter()
bootstats(1:5, smooth=function(b=b) jitter(b))

## Alternative entropy based smooth function
x <- round(rnorm(15, 10, 2))

entropy <- function(x, base=2) {
    freqs <- table(x) / length(x)
    -sum(freqs * log(freqs, base=base))
}

H5 <- entropy(x, base=5)

bootstats(x, 
  smooth=function(b=b, x=x, H5=H5) {
      b + rnorm(length(b), 0, H5/sqrt(length(x)))
  })

bootstats(x)

## Return resamples and plot distribution of means
bst <- bootstats(c(1, 2, 2, 4), return_resamples=TRUE)
t(bst[[2]])
plot(density(colMeans(bst[[2]])))

AkselA/R-confintplot documentation built on May 9, 2019, 12:52 p.m.