Material for Sigrist (2023) - "A Comparison of Machine Learning Methods for Data with High-Cardinality Categorical Variables"

This repository contains material for reproducing the results of Sigrist (2023) - "A Comparison of Machine Learning Methods for Data with High-Cardinality Categorical Variables".

• ETL: code for preparing the data with instructions on where to download the data
• data: pre-processed data sets for modeling when the license of the original source permits it
• run_experiments.R: code for running the experiments
• results: raw results
• tune_pars: chosen tuning parameters
• cv_folds: sample splits when doing cross-validation

See Sigrist (2023) for more information.