Compute Feature Selection Using Repeated Boruta Algorithm

Repeatedly applies the Boruta feature selection algorithm and aggregates results to determine consistently selected features.

Usage

feature.selection.boruta(
  data,
  iterations = NULL,
  fix,
  doParallel = F,
  workers = NULL,
  file_name = NULL,
  threshold = NULL,
  return
)

Arguments

data

A data frame with the column "target" (factor) as the response and other columns as features.

iterations

Integer. The number of Boruta iterations to perform.

fix

Logical. If TRUE, applies TentativeRoughFix() to resolve tentative features after each iteration.

doParallel

Logical. Whether to use parallel processing.

workers

Integer. Number of CPU cores to use for parallel execution. If NULL, uses all available cores minus one.

file_name

A string for naming output plots and CSV files saved in the "Results/" directory.

threshold

A numeric value between 0 and 1. A feature must be confirmed in more than threshold * iterations to be finally labeled as confirmed.

return

Logical. Whether to save the resulting plots in the "Results/" directory.

Value

A list containing:

• A vector of confirmed features.

• A vector of tentative features.

• A data frame with median importance values and final decisions.