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