list_hyperparameter: Create Hyperparameter List from a leaderboard dataset

In HMDA: Holistic Multimodel Domain Analysis for Exploratory Machine Learning

Create Hyperparameter List from a leaderboard dataset

Description

Detects columns in a data frame that contain hyperparameters for H2O DRF/GBM algorithms and returns a list with the unique values from each parameter column.

Usage

list_hyperparameter(df)

Arguments

df	A data frame containing model results with hyperparameter columns.

Details

This function scans the column names of the input data frame for common H2O hyperparameter names, such as "ntrees", "max_depth", "min_rows", "sample_rate", "col_sample_rate_per_tree", "min_split_improvement", "learn_rate", "mtries", and "seed". It extracts the unique values from each matching column and returns them in a list. The resulting list can be used as a hyperparameter grid for tuning via H2O grid search functions.

Value

A named list where each hyperparameter element is a vector of unique values for a hyperparameter.

Author(s)

E. F. Haghish

HMDA documentation built on April 4, 2025, 6:06 a.m.