com.kpro.algorithm
Class Conclusion_Simple

java.lang.Object
  extended by com.kpro.algorithm.ConclusionAlgorithm
      extended by com.kpro.algorithm.Conclusion_Simple

public class Conclusion_Simple
extends ConclusionAlgorithm

a very simple conclusion class. result is based on the closest objects only, as determined by the sum of inverse distances of the accepted versus rejected policies. confidences is the ratio of sum inverse distances of the chosen decision, versus the sum of all inverse distances.

Version:
29.09.11.1
Author:
ngerstle

Field Summary
 
Fields inherited from class com.kpro.algorithm.ConclusionAlgorithm
distanceMetric, extraArgs
 
Constructor Summary
Conclusion_Simple(DistanceMetric dm, java.lang.String[] extraArgs)
           
 
Method Summary
 Action conclude(PolicyObject np, java.lang.Iterable<PolicyObject> releventSet)
          makes a decision on the reduced set.
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Constructor Detail

Conclusion_Simple

public Conclusion_Simple(DistanceMetric dm,
                         java.lang.String[] extraArgs)
Method Detail

conclude

public Action conclude(PolicyObject np,
                       java.lang.Iterable<PolicyObject> releventSet)
makes a decision on the reduced set. This class creates two lists, one for accepted policies and one for rejected. Assuming there are policies in both (easy decision otherwise), whether the policy is accepted or not will depend on the difference between the sum of inverse distances of the list items (excluding zero-distances), with the smaller sum indicating the more relevent decision.

Specified by:
conclude in class ConclusionAlgorithm
Parameters:
np - the object under consideration
releventSet - the reduced set of neighbors
Returns:
an arraylist of {Action a, double Confidence)