Influence Plot Using Jackknife-After-Bootstrap

DESCRIPTION:

Produces a plot, analogous to a Cook's distance plot, displaying the relative influence of each observation upon the functional under consideration when using jackknife-after-bootstrap.

USAGE:

plot.jack.after.bootstrap(x, nrow=NULL, grid.layout=T, id.outliers=T, threshold=x$threshold) 

REQUIRED ARGUMENTS:

x
object of class jack.after.bootstrap.

OPTIONAL ARGUMENTS:

nrow
number of rows of plots on the screen. If nrow is NULL and x has a dim.obs component, it is used to determine nrow. Otherwise, two rows are used unless there is only a single parameter.
grid.layout
logical flag indicating whether to display plots in a single grid or as separate plots when the resampled statistic is a vector.
id.outliers
logical flag indicating whether observation numbers should be printed for relative influences greater than the specified threshold.
threshold
threshold at which points are considered to have a relatively large influence. A straight line is drawn at this level, and observations with a relative influence greater than this threshold are labeled if id.outliers=T.

VALUE:

no value is returned.

SIDE EFFECTS:

For each column of x$rel.influence, a plot is produced in which high-density lines are used to display the relative influences. A solid horizontal line is plotted at the value of threshold. If id.outliers=T, then values greater than threshold are labeled with their observation number.

SEE ALSO:

, .

EXAMPLES:

temp <- bootstrap(stack.loss,var) 
temp.jab <- jack.after.bootstrap(temp) 
print(temp.jab) 
plot(temp.jab)