Dynamic Sequential Multivariable Multinomial Logistic Regression — run_multivariable_multinomial_logistic

Fits sequential multinomial logistic regression models using nnet::multinom. Implements an "Anti-Cartesian" row-indexing strategy to ensure perfect alignment of multi-level categorical outcomes across adjustment layers.

Usage

run_multivariable_multinomial_logistic_regression(
  data,
  outcomes,
  predictors,
  models_list,
  ref_level = NULL,
  output_dir = NULL,
  save_format = c("none", "docx", "csv", "txt")
)

Arguments

data: A data frame containing the variables.
outcomes: Character vector of multinomial outcome variables.
predictors: Character vector of primary predictors.
models_list: A NAMED list defining sequential adjustment layers.
ref_level: Optional string to set the reference level of the outcome.
output_dir: Optional path to save output files.
save_format: Save format: "none", "docx", "csv", or "txt".

Value

A list containing results data frame and saved file paths.

Examples

if (FALSE) { # \dontrun{
# Example with 3-level categorical outcome
library(nnet)
df_test <- survival::colon %>%
  dplyr::mutate(extent = factor(extent, labels = c("Sub", "Mus", "Ser", "Con")))

run_multivariable_multinomial_logistic_regression(
  data = df_test,
  outcomes = "extent",
  predictors = "sex",
  models_list = list("Basic" = c("age"), "Clinical" = c("obstruct")),
  save_format = "docx"
)
} # }