Package de.jungblut.classification.knn

Class SparseKNearestNeighbours

java.lang.Object

de.jungblut.classification.AbstractPredictor

de.jungblut.classification.AbstractClassifier

de.jungblut.classification.knn.AbstractKNearestNeighbours

de.jungblut.classification.knn.SparseKNearestNeighbours

All Implemented Interfaces:

Classifier, Predictor

public final class SparseKNearestNeighbours
extends AbstractKNearestNeighbours

K nearest neighbour classification algorithm that is seeded with a "database" of known examples and predicts based on the k-nearest neighbours majority vote for a class. An Inverted Index is used internally to speedup the searches.
TODO maybe we can add a sampling facility for larger data.

Field Summary

Fields inherited from class de.jungblut.classification.knn.AbstractKNearestNeighbours

k, numOutcomes

Constructor Summary

Constructors

Constructor	Description
SparseKNearestNeighbours(int numOutcomes, int k, DistanceMeasurer measurer)	Constructs a new knn classifier.

Method Summary

Modifier and Type	Method	Description
protected java.util.List<de.jungblut.jrpt.VectorDistanceTuple<de.jungblut.math.DoubleVector>>	getNearestNeighbours(de.jungblut.math.DoubleVector feature, int k)	Find the k nearest neighbours for the given feature.
void	train(java.lang.Iterable<de.jungblut.math.DoubleVector> features, java.lang.Iterable<de.jungblut.math.DoubleVector> outcome)	Trains this classifier with the given features and the outcome.

Methods inherited from class de.jungblut.classification.knn.AbstractKNearestNeighbours

predict, predictProbability

Methods inherited from class de.jungblut.classification.AbstractClassifier

train

Methods inherited from class de.jungblut.classification.AbstractPredictor

extractPredictedClass, extractPredictedClass, predictedClass, predictedClass

Methods inherited from class java.lang.Object

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

Methods inherited from interface de.jungblut.classification.Predictor

asClassifier, extractPredictedClass, extractPredictedClass, predictedClass, predictedClass

Constructor Detail

SparseKNearestNeighbours

public SparseKNearestNeighbours(int numOutcomes,
                                int k,
                                DistanceMeasurer measurer)

Constructs a new knn classifier.

Parameters:

numOutcomes - the number of different outcomes that can be predicted.

k - the number of neighbours to analyse to get a prediction (it does so by majority voting).

measurer - the distance measurer to use.

Method Detail

train

public void train(java.lang.Iterable<de.jungblut.math.DoubleVector> features,
                  java.lang.Iterable<de.jungblut.math.DoubleVector> outcome)

Description copied from interface: Classifier

Trains this classifier with the given features and the outcome. This is the streaming method for training, it takes parallel iterables.

Specified by:

train in interface Classifier

Overrides:

train in class AbstractClassifier

outcome - the outcome must have classes labeled as doubles. E.G. in the binary case you have a single element and decide between 0d and 1d. In higher dimensional cases you have each of these single elements mapped to a dimension.

getNearestNeighbours

protected java.util.List<de.jungblut.jrpt.VectorDistanceTuple<de.jungblut.math.DoubleVector>> getNearestNeighbours(de.jungblut.math.DoubleVector feature,
                                                                                                                   int k)

Description copied from class: AbstractKNearestNeighbours

Find the k nearest neighbours for the given feature.

Specified by:

getNearestNeighbours in class AbstractKNearestNeighbours

Parameters:

feature - the feature to find neighbours for.

k - the number of neighbours to find.

Returns:

a list of VectorDistanceTuple's that contain the outcome of the retrieved vectors.