Re: [R] Identity link in tweedie

From: Peter Dunn <dunn_at_usq.edu.au>
Date: Tue, 18 Dec 2007 08:38:52 +1000

Cristina

> I'm using the tweedie distribution package and I cant figure out
> how to run a model using an identity link. I know I can use a log
> link by having link.power=0 and I think identity would be
> link.power=1, but I'm not sure.

Yes, that is correct.

> Furthermore when I try running it
> with link.power=1 it requires starting values which I cant manage
> to give appropriately

Without any code to see, I will guess at what is happening. The trouble with using an identity link is that the predicted means, say mu, may be negative, where as the Tweedie distributions have mu greater than zero. In using the identity link, you may be asking R to fit the impossible.

Perhaps the question you need to ask is this: Why do you need an identity link?

P.

-- 
Dr Peter Dunn  |  dunn <at> usq.edu.au
Faculty of Sciences, USQ; http://www.sci.usq.edu.au/staff/dunn
Aust. Centre for Sustainable Catchments: www.usq.edu.au/acsc

This email (including any attached files) is confidentia...{{dropped:15}}

______________________________________________
R-help_at_r-project.org 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.
Received on Mon 17 Dec 2007 - 22:49:17 GMT

Archive maintained by Robert King, hosted by the discipline of statistics at the University of Newcastle, Australia.
Archive generated by hypermail 2.2.0, at Mon 17 Dec 2007 - 23:30:19 GMT.

Mailing list information is available at https://stat.ethz.ch/mailman/listinfo/r-help. Please read the posting guide before posting to the list.