control.presmooth {survPresmooth} | R Documentation |
Control Values for the presmooth() Function
Description
This function returns a list of values for control
parameters of the presmooth
function.
Usage
control.presmooth(n.boot = c(5000, 1000), q.weight = c(0.2, 0.8),
k = 1, length.grid.bw.plugin = 100, length.grid.ise = 100,
pilot.par.ini = NULL, save.data = FALSE, save.mise = FALSE,
na.action = na.omit)
Arguments
n.boot |
A numeric vector of length 2 or, alternatively, a numeric value specifying the number of bootstrap resamples used in bootstrap bandwidth selection. If a vector, the first element is used for S and H estimation, and the second for f and h. The default is c(5000, 1000). |
q.weight |
A numeric vector of length 2 specifying the order of quantiles of the observed times that determine the left- and right-ends of the support of the weight function. The default is c(0.2, 0.8). |
k |
A numeric value specifying the multiple of the data range used as the maximum possible value for the selected plug-in or bootstrap bandwidth. The default is 1. |
length.grid.bw.plugin |
An integer specifying the number of points of the grid used for numerical integration (Simpson's rule) of integrals involved in plug-in bandwidth selection. The default is 100. |
length.grid.ise |
An integer specifying the number of points of the grid used for computing the integrated squared error by numerical integration (Simpson's rule) in bootstrap bandwidth selection. The default is 100. |
pilot.par.ini |
A numeric vector of length 8 specifying the
initial values of optimization routines used for pilot bandwidth
computation in plug-in and bootstrap bandwidth selection (although, in
the last case, it is not needed for survival and cumulative hazard
function estimation). If |
save.data |
A logical value specifying if the data must be saved
as a component of the value returned by the |
save.mise |
A logical value specifying if the MISE computed for
bootstrap bandwidth selection must be saved as a component of the
value returned by the |
na.action |
A function specifying how to handle missing
values. The default value, na.omit, deletes the rows containing
missing values in the (internal) data frame formed by the |
Details
The default values of n.boot
represent a trade-off
between computational speed and precision in bootstrap bandwidth
selection. If enough computing power or time are available, it is
recommended to increase the number of bootstrap resamples up to 10000
(which could be very slow for hazard and density estimation).
The six first values of pilot.par.ini
are three pairs of
parameters specifying three Weibull distributions (in each pair, the
first element is the shape parameter; the second one, the scale
parameter). These three Weibull distributions form a mixture, the
weights of its first two components being given by the last two values
of pilot.par.ini
.
Value
A list whose components are the arguments of the function, its defaults being replaced with the values the function was called with.
Author(s)
Ignacio Lopez-de-Ullibarri [aut, cre], Maria Amalia Jacome [aut]