repicea.stats.estimates

Class BootstrapHybridPointEstimate

java.lang.Object

repicea.stats.RandomVariable<D>

repicea.stats.estimates.Estimate<UnknownDistribution>

repicea.stats.estimates.BootstrapHybridPointEstimate

All Implemented Interfaces:

java.io.Serializable, CentralMomentsGettable, NumberOfRealizationsProvider

public final class BootstrapHybridPointEstimate
extends Estimate<UnknownDistribution>
implements NumberOfRealizationsProvider

This class implements the bootstrap estimator of the total as developed by Fortin et al. (2018) in the context of hybrid inference. More specifically, it applies when (i) the variable of interest has not been observed but predicted by a model and (ii) the covariates are not censused but only observed in a sample of the population. The estimator takes into account the variance that stems from the model as well as that of the sampling design. IMPORTANT: The model must benefit from a full stochastic implementation.

See Also:

Fortin, M., Manso, R., and Schneider, R. 2018. Parametric bootstrap estimators for hybrid inference in forest inventories. Forestry 91(3): 354-365. , Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
class	BootstrapHybridPointEstimate.VariancePointEstimate

Nested classes/interfaces inherited from class repicea.stats.estimates.Estimate

Estimate.EstimatorType

Field Summary

Fields inherited from class repicea.stats.estimates.Estimate

estimatorType, rowIndex

Constructor Summary

Constructors

Constructor and Description
BootstrapHybridPointEstimate()

Method Summary

Modifier and Type	Method and Description
protected BootstrapHybridPointEstimate	add(BootstrapHybridPointEstimate estimate2) This method returns a MonteCarloEstimate instance that results from the sum of two MonteCarloEstimate instances with the same number of realizations.
void	addBootstrapHybridEstimate(BootstrapHybridPointEstimate estimate) This method is useful for large simulations.
void	addPointEstimate(PointEstimate<?> estimate) This method adds a realization of the point estimate.
ConfidenceInterval	getConfidenceIntervalBounds(double oneMinusAlpha) This method returns the probability of getting a lower valueand upper bound of a confidence intervals at probability level 1 - alpha
Estimate<?>	getDifferenceEstimate(Estimate<?> estimate2) This method returns an difference estimate.
Matrix	getMean() This method returns the estimate of the total, which is the mean of the realized point estimates.
int	getNumberOfRealizations() This method returns the number of realizations the estimate is based on
Estimate<?>	getProductEstimate(double scalar) This method returns an difference estimate.
protected Matrix	getQuantileForProbability(double probability)
int	getSampleSize() Returns the sample size the individual point estimates are based on or -1 if no point estimate is available at the moment.
Estimate<?>	getSumEstimate(Estimate<?> estimate2) This method returns a sum of two estimates.
BootstrapHybridPointEstimate.VariancePointEstimate	getUncorrectedVariance() This method returns the uncorrected variance of the total estimate.
Matrix	getVariance() This method returns the second central moment, i.e.
BootstrapHybridPointEstimate.VariancePointEstimate	getVarianceOfTotalEstimate() This method returns the corrected variance of the total estimate.
protected boolean	isMergeableEstimate(Estimate<?> estimate) This method checks if the two point estimates are compatible.
protected BootstrapHybridPointEstimate	multiply(double scalar) This method returns a MonteCarloEstimate instance that results from the product of original MonteCarloEstimate instance and a scalar.
protected BootstrapHybridPointEstimate	subtract(BootstrapHybridPointEstimate estimate2) This method returns a MonteCarloEstimate instance that results from the subtraction of two MonteCarloEstimate instances with the same number of realizations.

Methods inherited from class repicea.stats.estimates.Estimate

getEstimatorType, getProductEstimate, getProductOfManyEstimates, getRandomDeviate, getRowIndex, setRowIndex

Methods inherited from class repicea.stats.RandomVariable

getDistribution

Methods inherited from class java.lang.Object

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

Constructor Detail

BootstrapHybridPointEstimate

public BootstrapHybridPointEstimate()

Method Detail

addPointEstimate

public void addPointEstimate(PointEstimate<?> estimate)

This method adds a realization of the point estimate. The compatibility of the instance with previously added instances is checked. If the check fails an InvalidParameterException is thrown.

Parameters:

estimate - a PointEstimate instance

isMergeableEstimate

protected boolean isMergeableEstimate(Estimate<?> estimate)

Description copied from class: Estimate

This method checks if the two point estimates are compatible. The basic check consists of comparing the classes. Then, the matrix data is checked for consistency with previous data.

Overrides:

isMergeableEstimate in class Estimate<UnknownDistribution>

Returns:

a boolean

addBootstrapHybridEstimate

public void addBootstrapHybridEstimate(BootstrapHybridPointEstimate estimate)

This method is useful for large simulations. The simulation can be run in different threads or batches and the resulting BootstrapHybridPointEstimate can then be combined through this method.

Parameters:

estimate - a BootstrapHybridPointEstimate instance that relies on the same PointEstimate class in the estimates member

getMean

public Matrix getMean()

This method returns the estimate of the total, which is the mean of the realized point estimates.

Specified by:

getMean in interface CentralMomentsGettable

Overrides:

getMean in class RandomVariable<UnknownDistribution>

Returns:

a Matrix instance

getUncorrectedVariance

public final BootstrapHybridPointEstimate.VariancePointEstimate getUncorrectedVariance()

This method returns the uncorrected variance of the total estimate. This estimator is based on the law of total variance. It tends to overestimate the true variance.

Returns:

a Matrix

See Also:

Fortin, M., Manso, R., and Schneider, R. 2018. Parametric bootstrap estimators for hybrid inference in forest inventories. Forestry 91(3): 354-365.

getVarianceOfTotalEstimate

public final BootstrapHybridPointEstimate.VariancePointEstimate getVarianceOfTotalEstimate()

This method returns the corrected variance of the total estimate. This estimator is unbiased.

Returns:

a VariancePointEstimate

See Also:

Fortin, M., Manso, R., and Schneider, R. 2018. Parametric bootstrap estimators for hybrid inference in forest inventories. Forestry 91(3): 354-365.

getVariance

public Matrix getVariance()

Description copied from interface: CentralMomentsGettable

This method returns the second central moment, i.e. the variance-covariance matrix, of the random variable.

Specified by:

getVariance in interface CentralMomentsGettable

Overrides:

getVariance in class RandomVariable<UnknownDistribution>

Returns:

a Number instance

getNumberOfRealizations

public int getNumberOfRealizations()

Description copied from interface: NumberOfRealizationsProvider

This method returns the number of realizations the estimate is based on

Specified by:

getNumberOfRealizations in interface NumberOfRealizationsProvider

Returns:

an integer

getQuantileForProbability

protected Matrix getQuantileForProbability(double probability)

getConfidenceIntervalBounds

public ConfidenceInterval getConfidenceIntervalBounds(double oneMinusAlpha)

Description copied from class: Estimate

This method returns the probability of getting a lower valueand upper bound of a confidence intervals at probability level 1 - alpha

Specified by:

getConfidenceIntervalBounds in class Estimate<UnknownDistribution>

Parameters:

oneMinusAlpha - is 1 minus the probability of Type I error

Returns:

a ConfidenceInterval instance

getSampleSize

public int getSampleSize()

Returns the sample size the individual point estimates are based on or -1 if no point estimate is available at the moment.

Returns:

an integer

subtract

protected BootstrapHybridPointEstimate subtract(BootstrapHybridPointEstimate estimate2)

This method returns a MonteCarloEstimate instance that results from the subtraction of two MonteCarloEstimate instances with the same number of realizations.

Parameters:

estimate2 - the estimate that is subtracted from this estimate

Returns:

a BootstrapHybridPointEstimate instance

add

protected BootstrapHybridPointEstimate add(BootstrapHybridPointEstimate estimate2)

This method returns a MonteCarloEstimate instance that results from the sum of two MonteCarloEstimate instances with the same number of realizations.

Parameters:

estimate2 - the estimate that is added to this estimate

Returns:

a BootstrapHybridPointEstimate instance

multiply

protected BootstrapHybridPointEstimate multiply(double scalar)

This method returns a MonteCarloEstimate instance that results from the product of original MonteCarloEstimate instance and a scalar.

Parameters:

scalar - the multiplication factor

Returns:

a BootstrapHybridPointEstimate instance

getDifferenceEstimate

public Estimate<?> getDifferenceEstimate(Estimate<?> estimate2)

Description copied from class: Estimate

This method returns an difference estimate.

Overrides:

getDifferenceEstimate in class Estimate<UnknownDistribution>

Parameters:

estimate2 - an Estimate to be subtracted from this estimate.

Returns:

an Estimate

getSumEstimate

public Estimate<?> getSumEstimate(Estimate<?> estimate2)

Description copied from class: Estimate

This method returns a sum of two estimates.

Overrides:

getSumEstimate in class Estimate<UnknownDistribution>

Parameters:

estimate2 - an Estimate to be added to this estimate.

Returns:

an Estimate

getProductEstimate

public Estimate<?> getProductEstimate(double scalar)

Description copied from class: Estimate

This method returns an difference estimate.

Overrides:

getProductEstimate in class Estimate<UnknownDistribution>

Parameters:

scalar - a double to be multiplied by this estimate

Returns:

an Estimate