boot.crq {crq}R Documentation

Bootstrap for censored regresion quantile models

Description

This function returns an array of bootstrap estimates for censored regression quantile models.

Usage

 boot.crq(x, y, c, taus, R = 100, mboot, bmethod = "xy-pair", ...) 

Arguments

x The design matrix
y The response vector
c The censoring indicator
taus The vector of taus at which to evaluate.
R Number of bootstrap replications, defaults to 100.
mboot Subsample size, defaults to the original sample size.
bmethod method for the bootstrap, either "xy-pair" or "Bose".
... Other parameters to be passed to crq.fit.

Details

Resampling function for crq. If method = "xy-pair", the default, then R samples of size mboot are taken and crq estimates are computed. If method = "Bose" the generalized resampling strategy of drawing standard exponential weights is used.

Value

Returns an list of three components, the of dimension p+1 by length(taus) by R, where p is the row dimension of the design matrix, i.e. the dimension of the model parameter.

Author(s)

Stephen Portnoy & Tereza Neocleous

See Also

crq


[Package crq version 0.3 Index]