jsat.regression

Class OrdinaryKriging

java.lang.Object

jsat.regression.OrdinaryKriging

All Implemented Interfaces:

java.io.Serializable, java.lang.Cloneable, Parameterized, Regressor

public class OrdinaryKriging
extends java.lang.Object
implements Regressor, Parameterized

An implementation of Ordinary Kriging with support for a uniform error measurement. When an error value is applied, Kriging becomes equivalent to Gaussian Processes Regression.

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	OrdinaryKriging.PowVariogram
static interface	OrdinaryKriging.Variogram

Field Summary

Fields

Modifier and Type	Field and Description
static double	DEFAULT_ERROR The default error value is DEFAULT_ERROR
static double	DEFAULT_NUGGET The default nugget value is 0.1

Constructor Summary

Constructors

Constructor and Description
OrdinaryKriging() Creates a new Ordinary Kriging with a small error value using the power variogram.
OrdinaryKriging(OrdinaryKriging.Variogram vari) Creates a new Ordinary Kriging with a small error value
OrdinaryKriging(OrdinaryKriging.Variogram vari, double error) Creates a new Ordinary Kriging
OrdinaryKriging(OrdinaryKriging.Variogram vari, double error, double nugget) Creates a new Ordinary Kriging.

Method Summary

Modifier and Type	Method and Description
OrdinaryKriging	clone()
double	getMeasurementError() Returns the measurement error used for Kriging, which is equivalent to altering the diagonal values of the covariance.
double	getNugget() Returns the nugget value passed to the variogram during training.
double	regress(DataPoint data)
void	setMeasurementError(double error) Sets the measurement error used for Kriging, which is equivalent to altering the diagonal values of the covariance.
void	setNugget(double nugget) Sets the nugget value passed to the variogram during training.
boolean	supportsWeightedData()
void	train(RegressionDataSet dataSet, boolean parallel)

Methods inherited from class java.lang.Object

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

Methods inherited from interface jsat.regression.Regressor

train

Methods inherited from interface jsat.parameters.Parameterized

getParameter, getParameters

Field Detail

DEFAULT_NUGGET

public static final double DEFAULT_NUGGET

The default nugget value is 0.1

See Also:

Constant Field Values

DEFAULT_ERROR

public static final double DEFAULT_ERROR

The default error value is DEFAULT_ERROR

See Also:

Constant Field Values

Constructor Detail

OrdinaryKriging

public OrdinaryKriging(OrdinaryKriging.Variogram vari,
                       double error,
                       double nugget)

Creates a new Ordinary Kriging.

Parameters:

vari - the variogram to fit to the data

error - the global measurement error

nugget - the nugget value to add to the variogram

OrdinaryKriging

public OrdinaryKriging(OrdinaryKriging.Variogram vari,
                       double error)

Creates a new Ordinary Kriging

Parameters:

vari - the variogram to fit to the data

error - the global measurement error

OrdinaryKriging

public OrdinaryKriging(OrdinaryKriging.Variogram vari)

Creates a new Ordinary Kriging with a small error value

Parameters:

vari - the variogram to fit to the data

OrdinaryKriging

public OrdinaryKriging()

Creates a new Ordinary Kriging with a small error value using the power variogram.

Method Detail

regress

public double regress(DataPoint data)

Specified by:

regress in interface Regressor

train

public void train(RegressionDataSet dataSet,
                  boolean parallel)

Specified by:

train in interface Regressor

supportsWeightedData

public boolean supportsWeightedData()

Specified by:

supportsWeightedData in interface Regressor

clone

public OrdinaryKriging clone()

Specified by:

clone in interface Regressor

Overrides:

clone in class java.lang.Object

setMeasurementError

public void setMeasurementError(double error)

Sets the measurement error used for Kriging, which is equivalent to altering the diagonal values of the covariance. While the measurement errors could be per data point, this implementation provides only a global error. If the error is set to zero, it will perfectly interpolate all data points.
Increasing the error smooths the interpolation, and has a large impact on the regression results.

Parameters:

error - the measurement error for all data points

getMeasurementError

public double getMeasurementError()

Returns the measurement error used for Kriging, which is equivalent to altering the diagonal values of the covariance. While the measurement errors could be per data point, this implementation provides only a global error. If the error is set to zero, it will perfectly interpolate all data points.

Returns:

the global error used for the data

setNugget

public void setNugget(double nugget)

Sets the nugget value passed to the variogram during training. The nugget allows the variogram to start from a non-zero value, and is equivalent to alerting the off diagonal values of the covariance.
Altering the nugget value has only a minor impact on the output

Parameters:

nugget - the new nugget value

Throws:

java.lang.ArithmeticException - if a negative nugget value is provided

getNugget

public double getNugget()

Returns the nugget value passed to the variogram during training. The nugget allows the variogram to start from a non-zero value, and is equivalent to alerting the off diagonal values of the covariance.

Returns:

the nugget added to the variogram