TrainingModels.md

Training Interfaces

• Training Interfaces
  • For classification
  • For regression

All classification training classes provide the ITrainingModel interface; all regression training classes provide the IRegressionTrainingModel interface.

For classification

The ITrainingModel interface exposes a Train method that accepts the input data as an object implementing the IProblem interface and returns the model implementing the IModel interface.

class NEOML_API ITrainingModel {
public:
	virtual ~ITrainingModel() = 0;

	// Train a classifier on the provided input data
	virtual CPtr<IModel> Train( const IProblem& trainingClassificationData ) = 0;
};

For regression

The IRegressionTrainingModel interface exposes a TrainRegression method that accepts the input data as an object implementing the IRegressionProblem interface and returns the model implementing the IRegressionModel interface.

class NEOML_API IRegressionTrainingModel {
public:
	virtual ~RegressionITrainingModel() = 0;

	// Train a regression model using the provided input values
	virtual CPtr<IRegressionModel> TrainRegression( const IRegressionProblem& problem ) = 0;
};