`insight::get_parameters()` ignores `effects`, `component` and `parameters` arguments for nonlinear `brms` model

[Sandhu-SS]

Hi

The bayestestR::bayesfactor_parameters() applied to the brms linear model respects effects, component and parameters options that allows for computing bayes_factor for a single parameter.

However, for nonlinear model, these options are ignored, and results are computed for all parameters including the group level effects (in a sense option 'full').

This difference could be resulting from the insight::find_parameters()

Below, I first show the results for the linear fit and then the nonlinear

Linear model

library(brms)
library(bayestestR)

contrasts(sleep$group) <- contr.equalprior_pairs # see vingette
my_custom_priors <-
  set_prior("student_t(3, 0, 1)", class = "b") +
  set_prior("student_t(3, 0, 1)", class = "sd", group = "ID")


brms_model <- suppressWarnings(brm(extra ~ group + (1 | ID),
                                   data = sleep,
                                   prior = my_custom_priors,
                                   refresh = 0
                                   ))

bayesfactor_parameters(brms_model, 
                                   effects = "fixed",
                                   parameters = "b_Intercept",
                                   verbose = T)
#> Sampling priors, please wait...
#> Warning: Bayes factors might not be precise.
#>   For precise Bayes factors, sampling at least 40,000 posterior samples is
#>   recommended.
#> Bayes Factor (Savage-Dickey density ratio)
#> 
#> Parameter   |   BF
#> ------------------
#> (Intercept) | 5.72
#> 
#> * Evidence Against The Null: 0
Created on 2025-05-26 with [reprex v2.1.1](https://reprex.tidyverse.org/)

Nonlinear model

library(brms)
library(bayestestR)
  
brms_nl <- fit_loss <- brm(
  bf(cum ~ ult * (1 - exp(-(dev/theta)^omega)),
     ult ~ 1 + (1|AY), omega ~ 1, theta ~ 1,
     nl = TRUE),
  data = brms::loss, family = gaussian(),
  prior = c(
    prior(normal(5000, 1000), nlpar = "ult"),
    prior(normal(1, 2), nlpar = "omega"),
    prior(normal(45, 10), nlpar = "theta")
  ),
  control = list(adapt_delta = 0.9),
  refresh = 0
)

bayestestR::bayesfactor_parameters(brms_nl, 
                                   effects = "fixed", 
                                   parameters = "b_ult_Intercept",
                                   verbose = T)
#> Sampling priors, please wait...
#> Warning: Bayes factors might not be precise.
#>   For precise Bayes factors, sampling at least 40,000 posterior samples is
#>   recommended.
#> Bayes Factor (Savage-Dickey density ratio) (sigma)
#> 
#> Parameter |       BF
#> --------------------
#> sigma     | 2.77e+10
#> 
#> # Fixed effects (ult)
#> 
#> Parameter   |       BF
#> ----------------------
#> (Intercept) | 8.59e+11
#> 
#> # Fixed effects (omega)
#> 
#> Parameter   |       BF
#> ----------------------
#> (Intercept) | 1.01e+27
#> 
#> # Fixed effects (theta)
#> 
#> Parameter   |       BF
#> ----------------------
#> (Intercept) | 5.17e+30
#> 
#> # Random effects (ult) (Intercept: AY)
#> 
#> Parameter |     BF
#> ------------------
#> AY.1991   | 983.54
#> AY.1992   |  0.516
#> AY.1993   |  0.509
#> AY.1994   |  0.781
#> AY.1995   |   2.04
#> AY.1996   |  0.464
#> AY.1997   |   1.82
#> AY.1998   |  25.77
#> AY.1999   |   1.20
#> AY.2000   |   1.05
#> 
#> # Random effects (ult) (SD/Cor: AY)
#> 
#> Parameter   |       BF
#> ----------------------
#> (Intercept) | 8.57e+05
#> 
#> * Evidence Against The Null: 0
Created on 2025-05-26 with [reprex v2.1.1](https://reprex.tidyverse.org/)

[mattansb]

Thanks for the report - this is indeed related to insight::get_parameters().