de.lmu.ifi.dbs.elki.algorithm.statistics

Class EvaluateRetrievalPerformance.KNNEvaluator

java.lang.Object

de.lmu.ifi.dbs.elki.algorithm.statistics.EvaluateRetrievalPerformance.KNNEvaluator

Enclosing class:

EvaluateRetrievalPerformance<O>

public static class EvaluateRetrievalPerformance.KNNEvaluator
extends Object

Evaluate kNN retrieval performance.

Author:

Erich Schubert

Field Summary

Fields

Modifier and Type	Field and Description
static EvaluateRetrievalPerformance.KNNEvaluator	STATIC Static instance.

Constructor Summary

Constructors

Constructor and Description
KNNEvaluator()

Method Summary

Modifier and Type	Method and Description
int	countkNN(gnu.trove.map.hash.TObjectIntHashMap<Object> counters, Object l) Counting helper for kNN classification.
void	evaluateKNN(double[] knnperf, ModifiableDoubleDBIDList nlist, Relation<?> lrelation, gnu.trove.map.hash.TObjectIntHashMap<Object> counters, Object label) Evaluate by simulating kNN classification for k=1...maxk

Methods inherited from class java.lang.Object

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

Field Detail

STATIC

public static final EvaluateRetrievalPerformance.KNNEvaluator STATIC

Static instance.

Constructor Detail

KNNEvaluator

public KNNEvaluator()

Method Detail

evaluateKNN

public void evaluateKNN(double[] knnperf,
                        ModifiableDoubleDBIDList nlist,
                        Relation<?> lrelation,
                        gnu.trove.map.hash.TObjectIntHashMap<Object> counters,
                        Object label)

Evaluate by simulating kNN classification for k=1...maxk

Parameters:

knnperf - Output data storage

nlist - Neighbor list

lrelation - Label relation

counters - (Reused) map for counting the class occurrences.

label - Label(s) of query object

countkNN

public int countkNN(gnu.trove.map.hash.TObjectIntHashMap<Object> counters,
                    Object l)

Counting helper for kNN classification.

Parameters:

counters - Counter storage

l - Object labels

Returns:

Maximum count