edu.indiana.iucbrf.adaptation
Class WeightedMajorityAdapter
java.lang.Object
edu.indiana.iucbrf.adaptation.Adaptation
edu.indiana.iucbrf.adaptation.WeightedMajorityAdapter
- All Implemented Interfaces:
- java.io.Serializable
public class WeightedMajorityAdapter
- extends Adaptation
This class adapts solutions based on a weighted majority vote of each feature.
Values examined in the voting are those returned by the Solution class' doubleValue() method.
=== A note regarding weights in this framework ===
There are many different types of weights used in different parts of the framework:
- in adaptation.WeightedAverageAdapter and adaptation.WeightedMajorityAdapter - The adaptation weights in these classes determine how important each particular case is in the final adaptation.
- in adaptation.DistanceWeightedAdapter - Similar idea to WeightedAverageAdapter and WeightedMajorityAdapter, except that the weights change for each adaptation. The weight of a case is proportional to the similarity of that case to the current problem.
- in problem.Problem.weightedEuclideanDistance() - the weights sent to this method describe how important a particular feature is in determining similarity information.
- See Also:
- Serialized Form
Methods inherited from class java.lang.Object |
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait |
WeightedMajorityAdapter
public WeightedMajorityAdapter()
- Construct a WeightedAverageAdapter without specifying the weights. The weights must
be specified with setWeights() before the adapter is used.
WeightedMajorityAdapter
public WeightedMajorityAdapter(double[] wts)
- Creates new WeightedMajorityAdapter
- Parameters:
Wts
- An array of doubles, of weights to be used by the majority adapter.
WeightedMajorityAdapter
public WeightedMajorityAdapter(double[] wts,
boolean debugMode)
- Creates new WeightedMajorityAdapter
- Parameters:
Wts
- An array of doubles, of weights to be used by the majority adapter.debugMode
- Set to true to print debugging information, false otherwise.
adaptTo
public Solution adaptTo(RetrievalResults retrieved,
Problem currentProblem,
Domain domain)
- Adapt retrieved solutions by taking a weighted majority vote.
The retrieved solutions may be any subclasses of Solution that define doubleValue().
- Specified by:
adaptTo
in class Adaptation
- Parameters:
Retrieved
- A RetrievalResults of Case objects deemed relevant to this problemCurrentProblem
- The Problem object being adapted to. In this implementation, this parameter is ignored.
- Returns:
- A Solution object of the adapted solution.