ml_grid.pipeline.hierarchical_hyperparameter_search

Enhanced hyperparameter search with hierarchical optimization.

Integrates the HierarchicalSearchOptimizer with existing Grid/Random/Bayesian search.

Classes

HierarchicalHyperparameterSearch

Initialize hierarchical hyperparameter search.

Module Contents

class ml_grid.pipeline.hierarchical_hyperparameter_search.HierarchicalHyperparameterSearch(algorithm: Any, parameter_space: Dict[str, Any], method_name: str, global_params: Any, ml_grid_object: Any, max_total_evals: int = 100, reduction_factor: float = 0.4, eval_function: Callable | None = None)

Initialize hierarchical hyperparameter search.

Parameters:

• algorithm – The scikit-learn compatible estimator instance

• parameter_space – Hyperparameter search space dictionary

• method_name – Name of the algorithm for logging

• global_params – Global parameters singleton instance

• ml_grid_object – Main pipeline object containing data and settings

• max_total_evals – Total evaluation budget across all stages

• reduction_factor – Space reduction factor per stage (0-1)

• eval_function – Optional custom evaluation function Signature: params -> score, fit_time

algorithm

parameter_space

method_name

global_params

ml_grid_object

max_total_evals = 100

reduction_factor = 0.4

logger

eval_function = None

run_hierarchical_search(X_train: pandas.DataFrame, y_train: pandas.Series) → Tuple[Any, Dict[str, List[Dict]]]

Execute hierarchical hyperparameter search.

Parameters:

• X_train – Training features DataFrame

• y_train – Training labels Series

Returns:

Tuple of (best_estimator, all_results_by_stage)

The method performs: 1. Coarse search: Broad exploration with minimal evaluations per parameter 2. Fine search: Focused exploitation on promising regions 3. Refinement: Detailed optimization of top candidates

Each stage uses dynamic space reduction based on previous results.