aft.fun {rankreg} | R Documentation |
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.
aft.fun(x, y, delta, randomseed=10, weight="logrank", nstep=3, mcsize=100)
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. |
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.
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.
Original Splus code by Z. Jin. Adapted to R by Mai Zhou.
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)