Variance decomposition for brms models

var_decomp(brmsfit)

Arguments

brmsfit: The output of a brms model. You can use VarDecomp::brms_model() to produce a brmsfit.

Value

Returns a data frame with posterior estimates.

Examples


# Simulate data
md = tibble::tibble(
  group = factor(sample(1:10, 1000, replace = TRUE)),
  f_var = factor(sample(1:3, 1000, replace = TRUE)),
  n_var = rnorm(1000, mean = 0, sd = 1),
  resp = rnorm(1000, mean = 10, sd = 3))

# Run model
mod = brms_model(Response = "resp", 
                 FixedEffect = c("f_var","n_var"), 
                 RandomEffect = "group", 
                 Family = "gaussian", 
                 Data = md)
#> [1] "No problems so far 🙃"
#> Compiling Stan program...
#> Start sampling


# Variance decomposition
var_decomp(mod)
#> # A tibble: 2,000 × 12
#>    Intercept  f_var2  f_var3    n_var sigma   R2_n_var R2_f_var3 R2_f_var2
#>        <dbl>   <dbl>   <dbl>    <dbl> <dbl>      <dbl>     <dbl>     <dbl>
#>  1      10.0 -0.197  -0.146  -0.150    3.01 0.00239    0.000517  0.000903 
#>  2      10.3 -0.0535 -0.418   0.00927  3.20 0.00000802 0.00373   0.0000585
#>  3      10.1 -0.261  -0.0566  0.0223   2.98 0.0000539  0.0000799 0.00162  
#>  4      10.5 -0.141  -0.749  -0.116    2.98 0.00144    0.0138    0.000470 
#>  5      10.1 -0.0609  0.275  -0.0639   3.08 0.000415   0.00176   0.0000825
#>  6      10.1  0.193  -0.398  -0.203    2.95 0.00453    0.00400   0.000903 
#>  7      10.3 -0.0695 -0.744  -0.0897   3.03 0.000838   0.0132    0.000110 
#>  8      10.1 -0.180  -0.134  -0.188    2.95 0.00390    0.000455  0.000788 
#>  9      10.2 -0.414  -0.341  -0.0506   3.00 0.000273   0.00285   0.00401  
#> 10      10.0 -0.182  -0.317  -0.145    2.96 0.00228    0.00252   0.000790 
#> # ℹ 1,990 more rows
#> # ℹ 4 more variables: R2_FixedEffects <dbl>, R2_group <dbl>,
#> #   R2_RandomEffects <dbl>, R2_residual <dbl>