source: trunk/CrossPare/src/de/ugoe/cs/cpdp/dataprocessing/BiasedWeights.java @ 9

package de.ugoe.cs.cpdp.dataprocessing;

import org.apache.commons.collections4.list.SetUniqueList;

import weka.core.Instance;
import weka.core.Instances;

/**
 * Sets the bias of the weights of the training data. By using a bias of 0.5 (default value) the total weight of the positive instances (i.e.
 * fault-prone) is equal to the total weight of the negative instances (i.e. non-fault-prone). Otherwise the weights between the two will be 
 * distributed according to the bias, where <0.5 means in favor of the negative instances and >0.5 in favor of the positive instances. 
 * equal to the total weight of the test 
 * @author Steffen Herbold
 */
public class BiasedWeights implements IProcessesingStrategy, ISetWiseProcessingStrategy {

        /**
         * bias used for the weighting
         */
        private double bias = 0.5;
        
        
        /**
         * Sets the bias to be used for weighting.
         * @param parameters string with the bias
         */
        @Override
        public void setParameter(String parameters) {
                bias = Double.parseDouble(parameters);
        }

        /**
         * @see de.ugoe.cs.cpdp.dataprocessing.ProcessesingStrategy#apply(weka.core.Instances, weka.core.Instances)
         */
        @Override
        public void apply(Instances testdata, Instances traindata) {
                //setBiasedWeights(testdata);
                setBiasedWeights(traindata);
        }
        
        /**
         * @see de.ugoe.cs.cpdp.dataprocessing.SetWiseProcessingStrategy#apply(weka.core.Instances, org.apache.commons.collections4.list.SetUniqueList)
         */
        @Override
        public void apply(Instances testdata, SetUniqueList<Instances> traindataSet) {
                for( Instances traindata : traindataSet ) {
                        setBiasedWeights(traindata);
                }
        }
        
        /**
         * Helper method that sets the weights for a given data set.
         * @param data data set whose weights are set
         */
        private void setBiasedWeights(Instances data) {
                final int classIndex = data.classIndex();
                
                final int[] counts = data.attributeStats(classIndex).nominalCounts;
                
                final double weightNegatives = ((1-bias)*data.numInstances()) / counts[0];
                final double weightPositives = (bias*data.numInstances()) / counts[1];
                
                
                for( int i=0 ; i<data.numInstances() ; i++ ) {
                        Instance instance = data.instance(i);
                        if( instance.value(classIndex)==0 ) {
                                instance.setWeight(weightNegatives);
                        }
                        if( instance.value(classIndex)==1 ) {
                                instance.setWeight(weightPositives);
                        }
                }
        }

        

}