jcolibri.method.reuse.classification
Class SimilarityWeightedVotingMethod

java.lang.Object
  extended by jcolibri.method.reuse.classification.AbstractKNNClassificationMethod
      extended by jcolibri.method.reuse.classification.SimilarityWeightedVotingMethod
All Implemented Interfaces:
KNNClassificationMethod

public class SimilarityWeightedVotingMethod
extends AbstractKNNClassificationMethod

Provides the ability to classify a query by predicting its solution from supplied cases. Classification is done by similarity weighted voting, where each vote is based on the similarity of the case to the query. The class with the highest overall value is the predicted class.

Author:
Derek Bridge, Lisa Cummins 16/05/07

Constructor Summary
SimilarityWeightedVotingMethod()
           
 
Method Summary
 ClassificationSolution getPredictedSolution(java.util.Collection<RetrievalResult> cases)
          Predicts the class that has the highest value vote among the k most similar cases, where votes are based on similarity to the query.
 
Methods inherited from class jcolibri.method.reuse.classification.AbstractKNNClassificationMethod
getPredictedCase
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Constructor Detail

SimilarityWeightedVotingMethod

public SimilarityWeightedVotingMethod()
Method Detail

getPredictedSolution

public ClassificationSolution getPredictedSolution(java.util.Collection<RetrievalResult> cases)
Predicts the class that has the highest value vote among the k most similar cases, where votes are based on similarity to the query. If several classes receive the same highest vote, the class that has the lowest hash code is taken as the prediction.

Specified by:
getPredictedSolution in interface KNNClassificationMethod
Specified by:
getPredictedSolution in class AbstractKNNClassificationMethod
Parameters:
cases - an ordered list of cases along with similarity scores.
Returns:
Returns the predicted solution.

GAIA - Group for Artificial Intelligence Applications
http://gaia.fdi.ucm.es