[R] enquiry

[Vladimir Eremeev]

ah06981 wrote:
> 
> Dear R-help,
> 
> I am trying to estimate a Cox model with nested effects basing on the 
> minimization of the overall AIC; I have two frailties terms, both gamma 
> distributed.  There is a error message (theta2 argument  misses) and I 
> don?t understand why. I would like to know what I have wrong. Thank you 
> very much for your time.
> 
> 
> fitM7 <- coxph(Surv(lifespan,censured) ~ south + frailty(id, 
> dist='gamma')+ frailty(mob, dist='gamma'), data= usa)
> 
> 
> tempfun <- function(theta1, theta2) {
> 
> fit <- coxph(Surv(lifespan,censured) ~ south + frailty(id, 
> dist='gamma', sparse= TRUE, theta=theta1)+ frailty(mob, dist='gamma', 
> sparse =TRUE, theta=theta2), data=usa)
> aic <- (fit$loglik[2] - fit$loglik[1]) - sum(fit$df)
> return(2*aic)
> }
> nlminb(c(theta1=3.2,theta2=.2), tempfun)
> 
> Error in (get(temp))(x, ...) : argument "theta2" is missing, with no
> default
> 
> Best,
> Silvia
> 

You need to 
(1) either rewrite your function to accept a single argument and treat it as
a numeric vector:

tempfun <- function(thetas) {
theta1<-thetas[1]
theta2<-thetas[2]

<the rest of function code>
}

(2) or change the call to nlminb:
nlminb(3.2, tempfun,theta2=.2)

But, I suspect, the optimization in this case will be only on theta1.

-- 
View this message in context: http://www.nabble.com/enquiry-tf4428517.html#a12633376
Sent from the R help mailing list archive at Nabble.com.