org.dllearner.learningproblems
Class ScoreTwoValued

java.lang.Object
  org.dllearner.core.Score
      org.dllearner.learningproblems.ScorePosNeg
          org.dllearner.learningproblems.ScoreTwoValued

public class ScoreTwoValued
extends ScorePosNeg

Calculates accuracy and score (with respect to some length penalty) of a class description. TODO: In fact, a score value influencing a learning algorithm should not be calculated here, but rather in a separate heuristic as there are many methods to calculate such a value. This class should only be used for computing example coverage, accuracy etc.

Author:

Jens Lehmann

Constructor Summary

ScoreTwoValued(int conceptLength, double percentPerLengthUnit, Set<Individual> posAsPos, Set<Individual> posAsNeg, Set<Individual> negAsPos, Set<Individual> negAsNeg)

ScoreTwoValued(Set<Individual> posAsPos, Set<Individual> posAsNeg, Set<Individual> negAsPos, Set<Individual> negAsNeg)

Method Summary

double	getAccuracy() This method returns a value, which indicates how accurate a class description solves a learning problem.
Set<Individual>	getCoveredNegatives()
Set<Individual>	getCoveredPositives()
ScorePosNeg	getModifiedLengthScore(int newLength) The score of a concept depends on how good it classifies the examples of a learning problem and the length of the concept itself.
Set<Individual>	getNotCoveredNegatives()
Set<Individual>	getNotCoveredPositives()
double	getScoreValue() score = accuracy - 1 - length * length penalty
String	toString()

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

ScoreTwoValued

public ScoreTwoValued(Set<Individual> posAsPos,
                      Set<Individual> posAsNeg,
                      Set<Individual> negAsPos,
                      Set<Individual> negAsNeg)

ScoreTwoValued

public ScoreTwoValued(int conceptLength,
                      double percentPerLengthUnit,
                      Set<Individual> posAsPos,
                      Set<Individual> posAsNeg,
                      Set<Individual> negAsPos,
                      Set<Individual> negAsNeg)

Method Detail

getAccuracy

public double getAccuracy()

Description copied from class: Score

This method returns a value, which indicates how accurate a class description solves a learning problem.

Specified by:

getAccuracy in class Score

Returns:

A value between 0 and 1 indicating the quality (of a class description).

See Also:

LearningProblem.getAccuracy(Description)

getScoreValue

public double getScoreValue()

score = accuracy - 1 - length * length penalty

Specified by:

getScoreValue in class ScorePosNeg

toString

public String toString()

Overrides:

toString in class Object

getCoveredNegatives

public Set<Individual> getCoveredNegatives()

Specified by:

getCoveredNegatives in class ScorePosNeg

getCoveredPositives

public Set<Individual> getCoveredPositives()

Specified by:

getCoveredPositives in class ScorePosNeg

getNotCoveredPositives

public Set<Individual> getNotCoveredPositives()

Specified by:

getNotCoveredPositives in class ScorePosNeg

getNotCoveredNegatives

public Set<Individual> getNotCoveredNegatives()

Specified by:

getNotCoveredNegatives in class ScorePosNeg

getModifiedLengthScore

public ScorePosNeg getModifiedLengthScore(int newLength)

Description copied from class: ScorePosNeg

The score of a concept depends on how good it classifies the examples of a learning problem and the length of the concept itself. If a given concept is known to have equal classification properties than the concept this score object is based on, then this method can be used to calculate its score value by using the length of this concept as parameter.

Specified by:

getModifiedLengthScore in class ScorePosNeg

Parameters:

newLength - Length of the concept.

Returns:

Score.