print(x, digits=max(options()$digits-3,4), ..., graphical = x$graphical)
print
for printing tables.
TRUE
(the default if
x
is based on quantiles) most printing
is disabled.
x
invisibly.
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.
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