Configuration Objects

The estimators accept grouped configuration objects to keep advanced optimizer setups readable. The flat keyword parameters remain supported for backward compatibility, but new examples prefer these objects.

EvolutionConfig([population_size, ...])	Core genetic algorithm controls.
PopulationConfig(initializer, warm_start_configs)	Initial population strategy.
RuntimeConfig([n_jobs, pre_dispatch, ...])	Execution, parallelism, and result-collection controls.
OptimizationConfig([local_search, ...])	Optional quality controls used around the main GA loop.

class sklearn_genetic.EvolutionConfig(population_size: int = 50, generations: int = 80, crossover_probability: object = 0.8, mutation_probability: object = 0.1, tournament_size: int = 3, elitism: bool = True, keep_top_k: int = 1, criteria: str = 'max', algorithm: str = 'eaMuPlusLambda')

Core genetic algorithm controls.

class sklearn_genetic.PopulationConfig(initializer: str = 'smart', warm_start_configs: list[dict] = <factory>)

Initial population strategy.

class sklearn_genetic.RuntimeConfig(n_jobs: int | None = None, pre_dispatch: str | int | None = '2*n_jobs', error_score: float | str = nan, return_train_score: bool = False, use_cache: bool = True, parallel_backend: str = 'auto', verbose: bool = True)

Execution, parallelism, and result-collection controls.

class sklearn_genetic.OptimizationConfig(local_search: bool = False, local_search_top_k: int = 1, local_search_steps: int = 1, local_search_radius: float = 0.1, diversity_control: bool = True, diversity_threshold: float = 0.25, diversity_stagnation_generations: int = 5, diversity_mutation_boost: float = 2.0, random_immigrants_fraction: float = 0.1, adaptive_selection: bool = False, selection_pressure_min: int = 2, selection_pressure_max: int | None = None, offspring_diversity_retries: int = 0, fitness_sharing: bool = False, sharing_radius: float = 0.2, sharing_alpha: float = 1.0, final_selection: bool = False, final_selection_top_k: int = 3, final_selection_cv: object = None)

Optional quality controls used around the main GA loop.