Displaying 1 result from an estimated 1 matches for "wexb".
Did you mean:
web
2007 Dec 04
2
weighted Cox proportional hazards regression
....frame", row.names=1:24)
coxph( Surv(pseudotime, cc) ~ x + strata(riskset), weights=wt,
robust=T, method="breslow",data=d2)
I'm expecting a value of about 0.63 to 0.64 based on the data source
(simulated) and the following hand-coded MLE:
negloglik = function(beta,dat) {
dat$wexb = dat$wt * exp(dat$x * beta)
agged = aggregate(dat$wexb,list(riskset=dat$riskset),sum)
names(agged)[2] = "denom"
dat = merge(dat[dat$cc==1,],agged,by="riskset")
-sum(log(dat$wexb)-log(dat$denom))
}
nlm(negloglik,0,hessian=T,dat=d2)
Am I misunderstanding the meaning of...