List of functionals (summaries of a bootstrap distribution). These accept a matrix of bootstrap replicates as input, and possibly the observed value and weights, and calculate a summary such as quantiles, centered quantiles, standard error, mean or bias, for each column.
resampFunctionalList[["mean"]](x, weights, ...)
resampFunctionalList[["bias"]](x, observed, weights, ...)
resampFunctionalList[["se"]](x, weights, ...)
resampFunctionalList[["bias&se"]](x, observed, weights, ...)
resampFunctionalList[["mean&se"]](x, weights, ...)
resampFunctionalList[["quantiles"]](x, weights,
probs = c(0.025, 0.16, 0.5, 0.84, 0.975), ...)
resampFunctionalList[["centered quantiles"]](x, observed, weights,
probs = c(0.025, 0.16, 0.5, 0.84, 0.975), ...)
resampFunctionalList[["standardized quantiles"]](x, observed, weights,
probs = c(0.025, 0.16, 0.5, 0.84, 0.975), ...)
B rows (number of bootstrap samples) and
p columns
(length of the observed statistic), containing the bootstrap sample.
p of observed statistics. This argument is required
for functionals that use it.
B, or
NULL signifying equal weights.
Calculations are done for the weighted distributions.
colMeans for
"mean".
"mean",
"bias", and
"se"
examples return vectors of length
p,
the others return matrices with
p columns and two or five (by default)
columns.
These are used by and .
You may define your own functional to pass to those functions;
it should have the same initial arguments. The output need not have
p columns or length
p; however the plotting routines will work
best if it does. If the output has
names or
dimnames they
will be used in printing and plotting.
x <- qexp(ppoints(19)) boot <- bootstrap(x, mean) plot(boot) # slightly skewed jab <- jackknifeAfterBootstrap(boot) plot(jab) plot(jab, xaxis = "L") plot(jab, xaxis = "data") jab <- jackknifeAfterBootstrap(boot, functional = "Centered Quantiles") plot(jab) # See also extensive examples in the help files for jackknifeAfterBootstrap # and tiltAfterBootstrap