How can I use relaimpo with glm? I want to examine the relative importance of the predictors in my model.
I know the {relaimpo} package in R allows for the examination of relative importance for lm but my model a glm and doesn't have a Gaussian family.
Is there a package that I can use to examine relative importance in this situation?
Take for example the following glm:
glm(hp ~ vs + am + disp, family=poisson(), data=mtcars)

Is there a method that would allow me to examine the relative importance of the three predictors in this model either using {relaimpo} or some other software implementation?
 A: Yes, the package {domir} on CRAN can accommodate a glm.
GLM does not have an explained variance metric in quite the same way as lm but you can obtain a very similar result (i.e., an "lmg"-type metric) from it using the {pscl} package's pR2 function such as:
> domir::domin(hp ~ vs + am + disp, glm, list(pscl::pR2, "McFadden"), data = mtcars, family = poisson())
Overall Fit Statistic:      0.5997781 

General Dominance Statistics:
     General_Dominance Standardized Ranks
vs          0.24878939   0.41480242     2
am          0.03508896   0.05850324     3
disp        0.31589971   0.52669434     1

Conditional Dominance Statistics:
         IVs: 1     IVs: 2     IVs: 3
vs   0.46514343 0.24708946 0.03413528
am   0.05091885 0.03338903 0.02095901
disp 0.51211049 0.31419978 0.12138886

Complete Dominance Statistics:
           Dmned?vs Dmned?am Dmned?disp
Dmate?vs          0        1         -1
Dmate?am         -1        0         -1
Dmate?disp        1        1          0

The "General Dominance Statistics" produced by domir::domin are identical to relaimpo::calc.relimp's "lmg" type (see github page for an illustration).
