Print a Jackknife-After-Bootstrap Object

DESCRIPTION:

Print method for object produced by .

USAGE:

print(x, digits=max(options()$digits-3,4), ..., graphical = x$graphical) 

REQUIRED ARGUMENTS:

x
object produced by .

OPTIONAL ARGUMENTS:

digits
the number of significant digits to print.
...
additional arguments, passed to print for printing tables.
graphical
if TRUE (the default if x is based on quantiles) most printing is disabled.

VALUE:

returns x invisibly.

DETAILS:

If graphical=FALSE this prints a table (described below), correlations if present, a list of observations with absolute standardized influence greater than the specified tolerance, and possibly warnings about too-small sample sizes in the leave-out bootstrap distributions.

The table has one row for each dimension of a multi-dimensional functional, and columns for (1) the functional applied to the original bootstrap distribution, (2) the mean of the jackknife replicates (functional applied to the leave-out bootstrap distributions), and (3) jackknife estimates of the bias and standard error of the functional.

The jackknife estimate should be used with caution - any noise from random bootstrap sampling is multiplied by a factor of n by the jackknife estimates, which often renders these estimates unusable.

SEE ALSO:

.

EXAMPLES:

x <- qexp(ppoints(19))  # artificial skewed data 
boot <- bootstrap(x, mean) 
jab1 <- jackknifeAfterBootstrap(boot, control = "none") 
plot(jab1)  # note the noise 
jab1 # tables are not printed 
print(jab1, graphical = FALSE) # The bias and SE numbers are useless 
 
jab2 <- jackknifeAfterBootstrap(boot, "SE") # uses control 
plot(jab) 
print(jab2) 
 
# See more extensive examples in the help file for jackknifeAfterBootstrap