mlrl.testbed_sklearn.experiments package¶

Provides classes that allow running experiments using the scikit-learn framework.

class mlrl.testbed_sklearn.experiments.SkLearnClassificationProblem(base_learner: sklearn.base.BaseEstimator, prediction_type: PredictionType, predictor_factory: PredictorFactory, fit_kwargs: dict[str, Any] | None = None, predict_kwargs: dict[str, Any] | None = None)¶

Represents a classification problem to be tackled via the scikit-learn framework.

class mlrl.testbed_sklearn.experiments.SkLearnExperiment(args: Namespace, initial_state: ExperimentState, dataset_splitter: DatasetSplitter, training_procedure: TrainingProcedure | None = None, prediction_procedure: PredictionProcedure | None = None)¶

An experiment that trains and evaluates a machine learning model using the scikit-learn framework.

class Builder(initial_state: ExperimentState, dataset_splitter: DatasetSplitter)¶

Allows to configure and create instances of the class SkLearnExperiment.

class PredictionProcedure(problem_domain: SkLearnProblem)¶

Allows to obtain predictions from a scikit-learn estimator.

predict(state: ExperimentState) → Generator[PredictionState, None, None]¶: See mlrl.testbed.experiments.experiment.Experiment.PredictionProcedure.predict()

class TrainingProcedure(base_learner: sklearn.base.BaseEstimator, fit_kwargs: dict[str, Any] | None = None)¶

Allows to fit a scikit-learn estimator to a training dataset.

train(learner: Any | None, parameters: dict[str, Any], dataset: Any) → TrainingState¶: See mlrl.testbed.experiments.experiment.Experiment.TrainingProcedure.train()

class mlrl.testbed_sklearn.experiments.SkLearnProblem(base_learner: sklearn.base.BaseEstimator, prediction_type: PredictionType, predictor_factory: PredictorFactory, fit_kwargs: dict[str, Any] | None = None, predict_kwargs: dict[str, Any] | None = None)¶

An abstract base class for all classes that represent a specific problem domain to be tackled via the scikit-learn framework.

class PredictorFactory¶

Bases: ABC

An abstract base class for all factories that allow to create instances of type Predictor.

abstractmethod create() → Predictor¶

Creates and returns a new object of type Predictor.

abstract property feature_dtype: Any¶: The data type to be used for features.

property learner_name: str¶: See mlrl.testbed.experiments.problem_domain.ProblemDomain.learner_name()

class mlrl.testbed_sklearn.experiments.SkLearnRegressionProblem(base_learner: sklearn.base.BaseEstimator, prediction_type: PredictionType, predictor_factory: PredictorFactory, fit_kwargs: dict[str, Any] | None = None, predict_kwargs: dict[str, Any] | None = None)¶

Represents a regression problem to be tackled via the scikit-learn framework.

Subpackages¶