how to compare glm parameter values

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

how to compare glm parameter values

Joost van Evert-2
Hi list,

I have trained a logistic regression model on three diverse features.
Now I want to compare these values to get an idea of their influence on
the result. For most features the summary function returns p-values that
equal zero, which makes the p-values unusable.

I have the intuition it is impossible to compare the relevance of the
features based on their parameter values as their ranges are quite
different.

Is it possible to compare the relevance of features on their parameters,
if their range differs?

Regards,

Joost

______________________________________________
[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: how to compare glm parameter values

Liaw, Andy
It's inappropriate to use the p-values associated with the coefficients for
that purpose.  They tell you how sure you can be, based on the data you
have, that they are not equal to zero.

Prof. Fox's `effects' package on CRAN would be a better place to look, I
think.  I believe there's an JSS article describing the package as well.

Andy

From: Joost van Evert

>
> Hi list,
>
> I have trained a logistic regression model on three diverse features.
> Now I want to compare these values to get an idea of their
> influence on
> the result. For most features the summary function returns
> p-values that
> equal zero, which makes the p-values unusable.
>
> I have the intuition it is impossible to compare the relevance of the
> features based on their parameter values as their ranges are quite
> different.
>
> Is it possible to compare the relevance of features on their
> parameters,
> if their range differs?
>
> Regards,
>
> Joost
>
> ______________________________________________
> [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
>
>

______________________________________________
[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