R help - Apr 2007 - Logistic/Cox regression: Parameter estimates directly from model matrix

Hi out there

Is there a way to get the estimated coefficients in a logistic / Cox
regression without having to specify a 'formula' but by only giving the
model matrix?

Example for Cox regression:

## predictors
n <- 50
q1 <- rnorm(n)
q2 <- rgamma(n, 2, 2)
Z <- cbind(q1, q2)

## response
ttf <- rexp(n)
tf <- round(runif(n))

## compute estimates
res <- coxph(Surv(ttf, tf) ~ q1 + q2)
r <- res$coef

My goal is to have a function

estFromModelMatrix <- function(tf, ttf, Z){

      /* do something meaningful using built-in functions */

      return(r)}

I have written such functions myself using LL - maximization from
scratch, but these are slower than the built-in functions. Since I
intend to do some simulations (where I specify the model matrix, but not
want to give a 'formula' manually for each simulation scenario), it
would be nice to have a function estFromModelMatrix().

I searched the help extensively, but did not find a way to do this.

Hope I was clear enough, any help is appreciated!
Kaspar Rufibach


-- 
______________________________________
Kaspar Rufibach
Department of Statistics -- Sequoia Hall
390 Serra Mall
Stanford University
Stanford, CA 94305-4065

mailto:kaspar.rufibach at stanford.edu
skype:kasparrufibach
http://www.stanford.edu/~kasparr

On 4/6/07, Kaspar Rufibach <kaspar.rufibach at stanford.edu>
wrote:
> Hi out there
>
> Is there a way to get the estimated coefficients in a logistic / Cox
> regression without having to specify a 'formula' but by only giving
the
> model matrix?
See 'coxreg.fit' in package 'eha'. Or 'glm.fit' for
logistic regression.

hth,

G?ran
>
> Example for Cox regression:
>
> ## predictors
> n <- 50
> q1 <- rnorm(n)
> q2 <- rgamma(n, 2, 2)
> Z <- cbind(q1, q2)
>
> ## response
> ttf <- rexp(n)
> tf <- round(runif(n))
>
> ## compute estimates
> res <- coxph(Surv(ttf, tf) ~ q1 + q2)
> r <- res$coef
>
> My goal is to have a function
>
> estFromModelMatrix <- function(tf, ttf, Z){
>
>       /* do something meaningful using built-in functions */
>
>       return(r)}
>
> I have written such functions myself using LL - maximization from
> scratch, but these are slower than the built-in functions. Since I
> intend to do some simulations (where I specify the model matrix, but not
> want to give a 'formula' manually for each simulation scenario), it
> would be nice to have a function estFromModelMatrix().
>
> I searched the help extensively, but did not find a way to do this.
>
> Hope I was clear enough, any help is appreciated!
> Kaspar Rufibach
>
>
> --
> ______________________________________
> Kaspar Rufibach
> Department of Statistics -- Sequoia Hall
> 390 Serra Mall
> Stanford University
> Stanford, CA 94305-4065
>
> mailto:kaspar.rufibach at stanford.edu
> skype:kasparrufibach
> http://www.stanford.edu/~kasparr
>
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

-- 
G?ran Brostr?m