org.encog.ml.model

Class EncogModel

java.lang.Object

org.encog.ml.model.EncogModel

public class EncogModel
extends java.lang.Object

Encog model is designed to allow you to easily swap between different model types and automatically normalize data. It is designed to work with a VersatileMLDataSet only.

Constructor Summary

Constructors

Constructor and Description
EncogModel(VersatileMLDataSet theDataset) Construct a model for the specified dataset.

Method Summary

Modifier and Type	Method and Description
double	calculateError(MLMethod method, MLDataSet data) Calculate the error for the given method and dataset.
MLMethod	createMethod() Create the selected method.
MLMethod	crossvalidate(int k, boolean shuffle) Crossvalidate and fit.
VersatileMLDataSet	getDataset()
java.util.List<ColumnDefinition>	getInputFeatures()
java.util.Map<java.lang.String,MethodConfig>	getMethodConfigurations()
java.util.List<ColumnDefinition>	getPredictedFeatures()
StatusReportable	getReport()
MatrixMLDataSet	getTrainingDataset()
MatrixMLDataSet	getValidationDataset()
void	holdBackValidation(double validationPercent, boolean shuffle, int seed) Specify a validation set to hold back.
void	selectMethod(VersatileMLDataSet dataset, java.lang.String methodType) Select the method to create.
void	selectMethod(VersatileMLDataSet dataset, java.lang.String methodType, java.lang.String methodArgs, java.lang.String trainingType, java.lang.String trainingArgs) Select the method to use.
void	selectTraining(VersatileMLDataSet dataset, java.lang.String trainingType, java.lang.String trainingArgs) Select the training to use.
void	selectTrainingType(VersatileMLDataSet dataset) Select the training type.
void	setReport(StatusReportable report)
void	setTrainingDataset(MatrixMLDataSet trainingDataset)
void	setValidationDataset(MatrixMLDataSet validationDataset)
MLRegression	train(boolean shuffle)
MLRegression	train(int iterations, boolean shuffle)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

EncogModel

public EncogModel(VersatileMLDataSet theDataset)

Construct a model for the specified dataset.

Parameters:

theDataset - The dataset.

Method Detail

getDataset

public VersatileMLDataSet getDataset()

Returns:

the dataset

getInputFeatures

public java.util.List<ColumnDefinition> getInputFeatures()

Returns:

the inputFeatures

getPredictedFeatures

public java.util.List<ColumnDefinition> getPredictedFeatures()

Returns:

the predictedFeatures

holdBackValidation

public void holdBackValidation(double validationPercent,
                               boolean shuffle,
                               int seed)

Specify a validation set to hold back.

Parameters:

validationPercent - The percent to use for validation.

shuffle - True to shuffle.

seed - The seed for random generation.

calculateError

public double calculateError(MLMethod method,
                             MLDataSet data)

Calculate the error for the given method and dataset.

Parameters:

method - The method to use.

data - The data to use.

Returns:

The error.

crossvalidate

public MLMethod crossvalidate(int k,
                              boolean shuffle)

Crossvalidate and fit.

Parameters:

k - The number of folds.

shuffle - True if we should shuffle.

Returns:

The trained method.

getTrainingDataset

public MatrixMLDataSet getTrainingDataset()

Returns:

the trainingDataset

setTrainingDataset

public void setTrainingDataset(MatrixMLDataSet trainingDataset)

Parameters:

trainingDataset - the trainingDataset to set

getValidationDataset

public MatrixMLDataSet getValidationDataset()

Returns:

the validationDataset

setValidationDataset

public void setValidationDataset(MatrixMLDataSet validationDataset)

Parameters:

validationDataset - the validationDataset to set

selectMethod

public void selectMethod(VersatileMLDataSet dataset,
                         java.lang.String methodType,
                         java.lang.String methodArgs,
                         java.lang.String trainingType,
                         java.lang.String trainingArgs)

Select the method to use.

Parameters:

dataset - The dataset.

methodType - The type of method.

methodArgs - The method arguments.

trainingType - The training type.

trainingArgs - The training arguments.

createMethod

public MLMethod createMethod()

Create the selected method.

Returns:

The created method.

selectMethod

public void selectMethod(VersatileMLDataSet dataset,
                         java.lang.String methodType)

Select the method to create.

Parameters:

dataset - The dataset.

methodType - The method type.

selectTrainingType

public void selectTrainingType(VersatileMLDataSet dataset)

Select the training type.

Parameters:

dataset - The dataset.

selectTraining

public void selectTraining(VersatileMLDataSet dataset,
                           java.lang.String trainingType,
                           java.lang.String trainingArgs)

Select the training to use.

Parameters:

dataset - The dataset.

trainingType - The type of training.

trainingArgs - The training arguments.

getMethodConfigurations

public java.util.Map<java.lang.String,MethodConfig> getMethodConfigurations()

Returns:

the methodConfigurations

getReport

public StatusReportable getReport()

Returns:

the report

setReport

public void setReport(StatusReportable report)

Parameters:

report - the report to set

train

public MLRegression train(boolean shuffle)

train

public MLRegression train(int iterations,
                          boolean shuffle)