prediction function for clogit model

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

prediction function for clogit model

Arne Jol
Dear R-Help,

I wonder if there is a prediction function for a clogit model which can be
used in the same way as the predict function for the multinom model.

In prediction('multinommodel',testset ...)  it is possible to predict the
class or the class probabilities for a testset. There is a predict function
for the coxph model but I cannot find an way to use this to predict the
classes (1,2 or 3) or the class probabilities (0.2, 0.3, 0.5) for example.

Can someone help me with this?

Regards,
Arne Jol

______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Reply | Threaded
Open this post in threaded view
|

Re: prediction function for clogit model

Thomas Lumley
On Thu, 16 Feb 2006, Arne Jol wrote:

> Dear R-Help,
>
> I wonder if there is a prediction function for a clogit model which can be
> used in the same way as the predict function for the multinom model.
>
> In prediction('multinommodel',testset ...)  it is possible to predict the
> class or the class probabilities for a testset. There is a predict function
> for the coxph model but I cannot find an way to use this to predict the
> classes (1,2 or 3) or the class probabilities (0.2, 0.3, 0.5) for example.
>

I don't think this is going to be possible. The point of conditional
logistic regression is that the probabilities depend on stratum parameters
that cannot be estimated accurately. The conditional likelihood removes
these parameters, but the resulting model does not contain enough
information to estimate probabilities.

  -thomas

______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html