rankaft {rankreg}R Documentation

Compute Rank estimator in right censored data AFT model

Description

Compute the Gehan and Logrank type rank regression estimators in the censored AFT model, using linear programming. This function is similar to aft.fun() except we strip away the re-sampling part in order to speedup things.

Usage

rankaft(x, y, delta)  

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 delta should be of length n.

Details

This program is memory hungry. Caution: at least 1G of RAM needed for sample size 1000; at least 512MB RAM for sample size 400.

We cut the re-sampling part (from aft.fun () ) to save computing time/memory, and concentrate on the bottleneck.

For statistical infernce, there are three options: (1) re-sampling method to estimate the var-cov matrix (available in aft.fun) (2) score type test available from function RankRegV() and (3) by empirical likelihood (see the reference).

Value

A list with beta which is the Gehan (betag) and Logrank type (betal) estimate rbinded together; and residuals.

Author(s)

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

References

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

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.

Zhou, M. (2005). Empirical likelihood analysis of the rank estimator for the censored AFT model. {em Biometrika}, {bf 92}, 492-498.

Examples

data(myeloma)
rankaft(x=cbind(myeloma[,3],myeloma[,4]),y=myeloma[,1],delta=myeloma[,2])

[Package rankreg version 0.2-2 Index]