Skip to contents

Builds a symbolized_model from a brmsfit object. v0.8 covers the Gaussian conditional submodel (identity link) with optional (1 | g) random intercepts. Other families and brms's distributional-parameter formulas (sigma ~ z, etc.) return capability errors via capability_check().

Usage

# S3 method for class 'brmsfit'
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.

Credible intervals

brms is Bayesian, so the "confidence band" carried in fixed_effects and interpretation is a 95% credible interval – the 2.5% and 97.5% posterior quantiles. excludes_zero is the indicator for whether the credible interval excludes zero (i.e. the posterior puts negligible mass on either side of zero). ci_method is set to "credible" so renderers can label the band correctly.

References

Bürkner, P.-C. (2017). brms: An R Package for Bayesian Multilevel Models Using Stan. Journal of Statistical Software, 80(1), 1-28.