Template Class BaggingEnsemble

Inheritance Relationships

Base Type

Class Documentation

template<typename ModelType, typename PredictionType>
class genif::BaggingEnsemble : public genif::Learner<std::vector<ModelType>, std::vector<PredictionType>>

Provides a generic Bagging ensemble, that randomly subsamples given datasets and outputs a vector of learned models and predictions.

Template Parameters

  • ModelType: The type of model, that the learner yields as result from calling fit.

  • PredictionType: The type of prediction, that the learner yields as result from calling either fitPredict and predict.

Public Functions

BaggingEnsemble(const Learner<ModelType, PredictionType> &baseLearner, unsigned int nModels = 100, unsigned int sampleSize = 256, unsigned int workerCount = 1, int seed = -1)

Constructs a new instance of BaggingEnsemble.

Parameters

  • baseLearner: The base learner, which serves as a prototype for subsequent learning efforts.

  • nModels: The number of models to fit.

  • sampleSize: The number of observations to draw to fit each model.

  • workerCount: The number of workers, which should fit models in parallel.

  • seed: Seed to use for random number generation (-1 defaults to sysclock seed). Pass an integer for constant result across multiple runs.

Learner<std::vector<ModelType>, std::vector<PredictionType>> &fit(const MatrixX &dataset) override

Fit nModels using the supplied dataset.

For fitting, this method takes a copy of the baseLearner property and calls fit using a previously drawn sample of dataset.

Return

A reference to the current BaggingEnsemble instance. The fitted models may be retrieved by calling the getModels() function.

Parameters

  • dataset: The dataset used to fit models.

std::vector<PredictionType> predict(const MatrixX &dataset) const override

Make predictions by using the set of models, which were previously learned with the fit method.

Internally, this method calls the predict method of the given base learner and calls predict with every learned model and the supplied dataset.

Return

A vector of predictions.

Parameters

  • dataset: The dataset to use for prediction.

std::vector<ModelType> getModel() const

Gathers the list of learned models, which were previously learned with the fit method.

Return

A list of models.

unsigned int getNumberOfModels() const

Returns the number of models, which should be fitted in this ensemble.

Return

As stated above.

unsigned int getActualNumberOfModels() const

Returns the number of models, which have been fitted in this ensemble.

Return

As stated above.