R help - Jul 2012 - Help needed to tackle multicollinearity problem in count data with the help of R

Dear everyone,

I'm student of Masters in Statistics (Actuarial) from Central
University of Rajasthan, India. I am doing a major project work as a
part of the degree. My major project deals with fitting a glm model
for the data of car insurance. I'm facing the problem of
multicollinearity for this data which is visible by the plotting of
data. But I'm not able to test it. In the case of linear models one
can use VIF but what about count data. Another problem is as the
multicollinearity is visible from the plot one solution is poisson
ridge regression. But how to do the programming for this in R. I
thought of making changes to "glm.fit" might help. But the codes are
beyond my understanding. I'll be glad if someone can help me in this
regard.


-- 
Regards
Umesh Khatri

Dear Umesh Khatri,

The vif() function in the car package will compute (generalized) variance
inflation factors for GLMs.

What (if anything) to do about collinearity isn't in my opinion an
answerable question without knowing the details of your research. In particular,
ridge regression is not a panacea for collinearity problems.

Best,
 John

------------------------------------------------
John Fox
Sen. William McMaster Prof. of Social Statistics
Department of Sociology
McMaster University
Hamilton, Ontario, Canada
http://socserv.mcmaster.ca/jfox/

On Wed, 11 Jul 2012 19:54:48 +0530
 umesh khatri <khatriumesh88 at gmail.com> wrote:
> Dear everyone,
> 
> I'm student of Masters in Statistics (Actuarial) from Central
> University of Rajasthan, India. I am doing a major project work as a
> part of the degree. My major project deals with fitting a glm model
> for the data of car insurance. I'm facing the problem of
> multicollinearity for this data which is visible by the plotting of
> data. But I'm not able to test it. In the case of linear models one
> can use VIF but what about count data. Another problem is as the
> multicollinearity is visible from the plot one solution is poisson
> ridge regression. But how to do the programming for this in R. I
> thought of making changes to "glm.fit" might help. But the codes
are
> beyond my understanding. I'll be glad if someone can help me in this
> regard.
> 
> 
> -- 
> Regards
> Umesh Khatri
> 
> ______________________________________________
> 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.