Skip to contents

Applications of multiple model layouts with `write_models()`

Definitive guide

Source: vignettes/usage/multiple-exec.Rmd
multiple-exec.Rmd

The Convenience of write_models()

The write_models() function is good at its job. It handles multiple <var_id> object layouts or mappers, and you can perform:

  1. Multiple executions
  2. ANOVA models
  3. Comparison between models (soon)

These mappers must share the same <STAT_FN> estimation method, else the execution will be aborted.

How to load

Just like this vignette, let us use the usual workflow of statim’s author in his daily basis on statistical analysis, where box is a main tool to handle code reusability and package imports.

box::use(
    statim[
        write_models, prepare, anova, conclude, via, 
        TTEST, CORTEST, LINEAR_REG, 
        x_by, rel, pairwise
    ]
)
iris |> 
    write_models(
        mod1 = rel(Sepal.Length, Petal.Width), 
        mod2 = pairwise(where(is.numeric))
    )

-- Models ---------------------------------------------------------------------- 

  mod1 : Sepal.Length ; Petal.Width
  mod2 : where(is.numeric)
mtcars |> 
    write_models(
        f1 = mpg ~ 1, 
        f2 = update(f1, ~ . + disp),
        f3 = update(f2, ~ . + hp),
        f4 = update(f3, ~ . + drat), 
        f5 = update(f4, ~ . + wt)
    ) |> 
    prepare(LINEAR_REG) |> 
    anova()

== ANOVA · F =================================================================== 

-- ANOVA Table -----------------------------------------------------------------

───────────────────────────────────────────────────────────
  model  res_df  deviance  df  dev_diff  f_value  p_value  
───────────────────────────────────────────────────────────
   f1      31    1126.047                                  
   f2      30    317.159   1   808.888   119.450  <0.001   
   f3      29    283.493   1    33.665    4.971    0.034   
   f4      28    253.346   1    30.148    4.452    0.044   
   f5      27    182.838   1    70.508   10.412   <0.001   
───────────────────────────────────────────────────────────