File statistics_provider_decomposable_dense.hpp

namespace boosting
class DenseDecomposableClassificationStatisticsProviderFactory : public IClassificationStatisticsProviderFactory
#include <statistics_provider_decomposable_dense.hpp>

Allows to create instances of the class IStatisticsProvider that can be used in classification problems and provide access to an object of type IDecomposableStatistics using dense data structures for storing the statistics.

Public Functions

DenseDecomposableClassificationStatisticsProviderFactory(std::unique_ptr<IDecomposableClassificationLossFactory> lossFactoryPtr, std::unique_ptr<IClassificationEvaluationMeasureFactory> evaluationMeasureFactoryPtr, std::unique_ptr<IDecomposableRuleEvaluationFactory> defaultRuleEvaluationFactoryPtr, std::unique_ptr<IDecomposableRuleEvaluationFactory> regularRuleEvaluationFactoryPtr, std::unique_ptr<IDecomposableRuleEvaluationFactory> pruningRuleEvaluationFactoryPtr, uint32 numThreads)
Parameters:

  • lossFactoryPtr – An unique pointer to an object of type IDecomposableClassificationLossFactory that allows to create implementations of the loss function that should be used for calculating gradients and Hessians

  • evaluationMeasureFactoryPtr – An unique pointer to an object of type IClassificationEvaluationMeasureFactory that allows to create implementations of the evaluation measure that should be used for assessing the quality of predictions

  • defaultRuleEvaluationFactoryPtr – An unique pointer to an object of type IDecomposableRuleEvaluationFactory that should be used for calculating the predictions, as well as corresponding quality scores, of the default rule

  • regularRuleEvaluationFactoryPtr – An unique pointer to an object of type IDecomposableRuleEvaluationFactory that should be used for calculating the predictions, as well as corresponding quality scores, of all remaining rules

  • pruningRuleEvaluationFactoryPtr – An unique pointer to an object of type IDecomposableRuleEvaluationFactory that should be used for calculating the predictions, as well as corresponding quality scores, when pruning rules

  • numThreads – The number of CPU threads to be used to calculate the initial statistics in parallel. Must be at least 1

std::unique_ptr<IStatisticsProvider> create(const CContiguousView<const uint8> &labelMatrix) const override

See also

IClassificationStatisticsProviderFactory::create

std::unique_ptr<IStatisticsProvider> create(const BinaryCsrView &labelMatrix) const override

See also

IClassificationStatisticsProviderFactory::create

Private Members

const std::unique_ptr<IDecomposableClassificationLossFactory> lossFactoryPtr_
const std::unique_ptr<IClassificationEvaluationMeasureFactory> evaluationMeasureFactoryPtr_
const std::unique_ptr<IDecomposableRuleEvaluationFactory> defaultRuleEvaluationFactoryPtr_
const std::unique_ptr<IDecomposableRuleEvaluationFactory> regularRuleEvaluationFactoryPtr_
const std::unique_ptr<IDecomposableRuleEvaluationFactory> pruningRuleEvaluationFactoryPtr_
const uint32 numThreads_
class DenseDecomposableRegressionStatisticsProviderFactory : public IRegressionStatisticsProviderFactory
#include <statistics_provider_decomposable_dense.hpp>

Allows to create instances of the class IStatisticsProvider that can be used in regression problems and provide access to an object of type IDecomposableStatistics using dense data structures for storing the statistics.

Public Functions

DenseDecomposableRegressionStatisticsProviderFactory(std::unique_ptr<IDecomposableRegressionLossFactory> lossFactoryPtr, std::unique_ptr<IRegressionEvaluationMeasureFactory> evaluationMeasureFactoryPtr, std::unique_ptr<IDecomposableRuleEvaluationFactory> defaultRuleEvaluationFactoryPtr, std::unique_ptr<IDecomposableRuleEvaluationFactory> regularRuleEvaluationFactoryPtr, std::unique_ptr<IDecomposableRuleEvaluationFactory> pruningRuleEvaluationFactoryPtr, uint32 numThreads)
Parameters:

  • lossFactoryPtr – An unique pointer to an object of type IDecomposableRegressionLossFactory that allows to create implementations of the loss function that should be used for calculating gradients and Hessians

  • evaluationMeasureFactoryPtr – An unique pointer to an object of type IRegressionEvaluationMeasureFactory that allows to create implementations of the evaluation measure that should be used for assessing the quality of predictions

  • defaultRuleEvaluationFactoryPtr – An unique pointer to an object of type IDecomposableRuleEvaluationFactory that should be used for calculating the predictions, as well as corresponding quality scores, of the default rule

  • regularRuleEvaluationFactoryPtr – An unique pointer to an object of type IDecomposableRuleEvaluationFactory that should be used for calculating the predictions, as well as corresponding quality scores, of all remaining rules

  • pruningRuleEvaluationFactoryPtr – An unique pointer to an object of type IDecomposableRuleEvaluationFactory that should be used for calculating the predictions, as well as corresponding quality scores, when pruning rules

  • numThreads – The number of CPU threads to be used to calculate the initial statistics in parallel. Must be at least 1

std::unique_ptr<IStatisticsProvider> create(const CContiguousView<const float32> &regressionMatrix) const override

See also

IRegressionStatisticsProviderFactory::create

std::unique_ptr<IStatisticsProvider> create(const CsrView<const float32> &regressionMatrix) const override

See also

IRegressionStatisticsProviderFactory::create

Private Members

const std::unique_ptr<IDecomposableRegressionLossFactory> lossFactoryPtr_
const std::unique_ptr<IRegressionEvaluationMeasureFactory> evaluationMeasureFactoryPtr_
const std::unique_ptr<IDecomposableRuleEvaluationFactory> defaultRuleEvaluationFactoryPtr_
const std::unique_ptr<IDecomposableRuleEvaluationFactory> regularRuleEvaluationFactoryPtr_
const std::unique_ptr<IDecomposableRuleEvaluationFactory> pruningRuleEvaluationFactoryPtr_
const uint32 numThreads_