rankaft {rankreg} | R Documentation |
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.
rankaft(x, y, delta)
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. |
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).
A list with beta
which is the Gehan (betag)
and Logrank type (betal) estimate
rbinded together; and residuals
.
Original Splus code by Jin Z. Adapted to R by Mai Zhou.
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.
data(myeloma) rankaft(x=cbind(myeloma[,3],myeloma[,4]),y=myeloma[,1],delta=myeloma[,2])