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 type rank estimation equation. Finally, it compute 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 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
estimates from re-sampling.
covw
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)