Symbolize a base R glm() fit (Gaussian / binomial / poisson, v0.10 first slice)

Builds a symbolized_model from a base R glm object. v0.10 covers Gaussian / binomial / poisson with their canonical links (identity / logit / log). Quasi-families, offsets, and family-specific weights are deferred via the capability registry.

Usage

# S3 method for class 'glm'
symbolize(fit, symbols = NULL, units = NULL, context = NULL, ...)

Arguments

fit

A fitted statistical model object.

symbols

Optional named character vector mapping variable names to user-supplied LaTeX symbols, e.g. c(body_mass = "W_i", temperature = "T_i").

units

Optional named character vector mapping variable names to units, e.g. c(body_mass = "g", temperature = "C").

context

Optional short character description of the model, e.g. "avian body-size location-scale model".

...

Reserved for method-specific extra arguments.

Value

A symbolized_model object.

Confidence intervals

CIs come from stats::confint.default(fit) — fast Wald intervals. (The default stats::confint() for a glm uses profile likelihood, which is slow; symbolize() uses the Wald form deliberately.) ci_method is labelled "wald".

References

McCullagh, P., & Nelder, J. A. (1989). Generalized Linear Models (2nd ed.). Chapman and Hall/CRC.