Extract coefficients in terms of the original levels of the coefficients rather than the coded variables.
Usage
rating_factors(
...,
model_data = NULL,
exposure = NULL,
exponentiate = TRUE,
signif_stars = FALSE,
round_exposure = 0
)Arguments
- ...
glm object(s) produced by
glm()- model_data
data.frame used to create glm object(s), this should only be specified in case the exposure is desired in the output, default value is NULL
- exposure
column in
model_datawith exposure, default value is NULL- exponentiate
logical indicating whether or not to exponentiate the coefficient estimates. Defaults to TRUE.
- signif_stars
show significance stars for p-values (defaults to TRUE)
- round_exposure
number of digits for exposure (defaults to 0)
Details
A fitted linear model has coefficients for the contrasts of the factor terms, usually one less in number than the number of levels. This function re-expresses the coefficients in the original coding. This function is adopted from dummy.coef(). Our adoption prints a data.frame as output. Use rating_factors_() for standard evaluation.
Examples
df <- MTPL2 |>
dplyr::mutate(dplyr::across(c(area), as.factor)) |>
dplyr::mutate(dplyr::across(c(area), ~biggest_reference(., exposure)))
mod1 <- glm(nclaims ~ area + premium, offset = log(exposure),
family = poisson(), data = df)
mod2 <- glm(nclaims ~ area, offset = log(exposure), family = poisson(),
data = df)
rating_factors(mod1, mod2, model_data = df, exposure = exposure)
#> risk_factor level est_mod1 est_mod2 exposure
#> 1 (Intercept) (Intercept) 0.1279538 0.1369930 NA
#> 2 area 0 0.5465581 0.5485629 13
#> 3 area 1 1.0000000 1.0000000 1066
#> 4 area 2 0.8728835 0.8739528 819
#> 5 area 3 1.0762334 1.0782596 765
#> 6 premium premium 1.0012348 NA NA