aft.fun {rankreg}R Documentation

Rank estimator in censored AFT model

Description

Use linear programming to solve the Gehan rank estimation equation for censored AFT model. Use iterated Gehan type solutions to solve the Logrank estimation equation. Finally, it computes the variance-covariance estimator for both rank regression estimators in the censored AFT model by re-sampling.

Usage

aft.fun(x, y, delta, randomseed=10, weight="logrank", nstep=3, mcsize=100) 

Arguments

x the design matrix, of size n by q.
y a vector containing the censored responses in the AFT model.
delta a vector of 1's and 0's. censoring indicator. 1(uncensor), 0(censored). Both y and d should be of length n.
randomseed
weight
nstep an integer. The number of iterations used to compute the logrank type estimator starting from the Gehan estimator.
mcsize number of resamples used to compute the variance estimator.

Details

For data sets with more than 400 observations, this function is slow. The reason is that it needs to solve linear programming problems of size n square. So 400 becomes 160000.

Value

A list with the following components. beta: first column is the Gehan estimator, the rest are logrank type estimators. betaw: the estimates from re-sampling. covw: variance-covariance estimator of beta from resampling.

Author(s)

Original Splus code by Z. Jin. Adapted to R by Mai Zhou.

References

Jin, Z., Lin, D.Y., Wei, L. J. and Ying, Z. (2003). Rank-based inference for the accelerated failure time model. {em Biometrika}, {bf 90}, 341-353.

Kalbfleisch, J. and Prentice, R. (2002) The Statistical Analysis of Failure Time Data. 2nd Ed. Wiley, New York. (In particular Chapter 7)


[Package rankreg version 0.2-2 Index]