In erikerhardt/erikmisc: Erik Erhardt's miscellaneous functions for solving complex data analysis workflows

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(erikmisc)

Best subset selection

The e_lm_bestsubset() function is a wrapper for leaps::regsubsets(). It takes a "full model" formula and returns the best nbest models of each model size (1 predictor, 2 predictors, etc.), in a table that's easy to review.

NOTE: This does not assess model fit! For that, see e_lm_diag_plots().

Data

Fit the best 3 models of each model size. For each column, a 1 or 0 indicates the variable is or is not in the model. The model SIZE is how many terms are in the model (not counting the intercept). Other model statistics are provided and the results are sorted by BIC (lowest BIC is preferred, best on top).

tab_best <-
  e_lm_bestsubset(
    form  = formula(mpg ~ cyl + disp + hp + gear)
  , dat   = datasets::mtcars
  , nbest = 3
  )

  ### consider these options if you temporarily need a wider output in your Rmd output
  # op <- options(); # saving old options
  # options(width=100) # setting command window output text width wider
tab_best |> print(n = Inf, width = Inf)
  # options(op); # reset (all) initial options

# print into RStudio viewer or a pretty table in Rmd output
# include "always_allow_html: yes" in Rmd yaml header
tab_best |> e_table_print(sw_kable_format  = c("simple", "kbl", "html", "latex", "doc")[1])

erikerhardt/erikmisc documentation built on April 17, 2025, 10:48 a.m.