Configuration Reference¶

All parameters accepted by OptunaSearchCV.

Basic Parameters¶

Parameter	Type	Default	Description
`estimator`	estimator	required	The Scikit-Learn estimator to tune
`param_distributions`	`dict[str, BaseDistribution]`	required	Mapping of parameter names to Optuna distributions
`n_trials`	`int`	`10`	Number of Optuna trials to run
`timeout`	`float \\| None`	`None`	Maximum seconds for the study. `None` means no limit
`cv`	`int \\| CV splitter \\| None`	`None`	Cross-validation strategy. `None` defaults to 5-fold
`n_jobs`	`int \\| None`	`None`	Parallel trial jobs. `-1` uses all CPUs. `None` means 1
`refit`	`bool \\| str`	`True`	Refit the best estimator on the full dataset. Pass a metric name when using multi-metric scoring
`verbose`	`int`	`0`	Verbosity level for Optuna logging

Parameter	Type	Default	Description
`sampler`	`Sampler \\| None`	`None`	Sampler wrapper. `None` uses Optuna's default (TPE)
`storage`	`Storage \\| None`	`None`	Storage wrapper for trial persistence
`callbacks`	`dict[str, Callback] \\| None`	`None`	Dictionary of callback wrappers invoked after each trial
`scoring`	`str \\| callable \\| list \\| dict \\| None`	`None`	Scoring metric(s). `None` uses the estimator's default scorer
`return_train_score`	`bool`	`False`	Include training fold scores in `cv_results_`
`error_score`	`float \\| "raise"`	`np.nan`	Value assigned on fit failure. `"raise"` stops the search
`study_name`	`str \\| None`	`None`	Name for the Optuna study

Attribute	Type	Description
`best_params_`	`dict`	Parameter setting that gave the best score
`best_score_`	`float`	Mean cross-validated score of the best trial
`best_index_`	`int`	Index of the best trial in `cv_results_`
`best_estimator_`	estimator	Estimator refitted on full data (when `refit=True`)
`cv_results_`	`dict`	Dictionary with per-trial results in Scikit-Learn format
`study_`	`optuna.Study`	The Optuna study object
`trials_`	`list[FrozenTrial]`	List of all completed trials