Compute Cross-Validation Accuracy for ML Models — compute_cv

This function extracts cross-validated accuracy values from a list of trained machine learning models, summarizes their median and standard deviation, and optionally plots a bar chart or selects base models for stacking.

Usage

compute_cv_accuracy(
  models,
  file_name = NULL,
  base_models = FALSE,
  return = TRUE
)

Arguments

models: A named list of trained ML models, each with a resample element containing cross-validated accuracy.
file_name: (Optional) Character string specifying the filename prefix for the saved accuracy plot (PDF format).
base_models: Logical. If TRUE, the function selects and returns base models using choose_base_models() for stacking.
return: Logical. If TRUE, the function saves a barplot of the model accuracy values in the Results/ directory.

Value

A list containing:

Accuracy: A data frame with the median and standard deviation of accuracy for each model.
Top_model: A character string naming the model with the highest median accuracy.
Base_models (optional): A character vector of selected base models if base_models = TRUE.

Details

This function assumes that each model in the list has a $resample component containing a column named Accuracy. It calculates the median and standard deviation of accuracy for each model and creates a barplot (if return = TRUE) with error bars.

If base_models = TRUE, it calls a helper function choose_base_models() to select models for use in stacking.