edu.cmu.minorthird.classify
Interface Instance

All Superinterfaces:

HasSubpopulationId, Visible

All Known Implementing Classes:

AbstractInstance, AugmentedInstance, Example, FeatureFactory.CompactInstance, HyperplaneInstance, InstanceFromPrediction, InstanceFromSequence, MaskedInstance, MultiExample, MutableInstance, SGMExample

public interface Instance
extends HasSubpopulationId, Visible

A single instance for a learner. This is basically a weighted set of features.

Author:

William Cohen

Method Summary

java.util.Iterator<Feature>	binaryFeatureIterator() Return an iterator over all binary features.
java.util.Iterator<Feature>	featureIterator() Return an iterator over all features
java.lang.Object	getSource() Get the underlying object used that this instance describes
java.lang.String	getSubpopulationId() Get the subpopulation from which this instance was drawn.
double	getWeight(Feature f) Get the weight assigned to a feature in this instance.
java.util.Iterator<Feature>	numericFeatureIterator() Return an iterator over all numeric features.
int	numFeatures() Return the number of all the features

Methods inherited from interface edu.cmu.minorthird.util.gui.Visible

toGUI

Method Detail

getWeight

double getWeight(Feature f)

Get the weight assigned to a feature in this instance.

binaryFeatureIterator

java.util.Iterator<Feature> binaryFeatureIterator()

Return an iterator over all binary features.

numericFeatureIterator

java.util.Iterator<Feature> numericFeatureIterator()

Return an iterator over all numeric features.

featureIterator

java.util.Iterator<Feature> featureIterator()

Return an iterator over all features

numFeatures

int numFeatures()

Return the number of all the features

getSource

java.lang.Object getSource()

Get the underlying object used that this instance describes

getSubpopulationId

java.lang.String getSubpopulationId()

Get the subpopulation from which this instance was drawn. A null id is considered to be a unique subpopulation---different from every other subpopulation, including other nulls. A subpopulation is a subset of the training data which is expected to contain additional regularities - for instance, pages from the same site, or spans from the same document. Testing routines may use subpopulation's to more correctly spit datasets into train/test subsets.

Specified by:

getSubpopulationId in interface HasSubpopulationId