mlda.labelsRelation.SCUMBLE.java Source code

Introduction

Here is the source code for mlda.labelsRelation.SCUMBLE.java

Source

/*
 * This file is part of the MLDA.
 *
 * (c)  Jose Maria Moyano Murillo
 *      Eva Lucrecia Gibaja Galindo
 *      Sebastian Ventura Soto <sventura@uco.es>
 *
 * For the full copyright and license information, please view the LICENSE
 * file that was distributed with this source code.
 */

package mlda.labelsRelation;

import mlda.base.MLDataMetric;
import mulan.data.InvalidDataFormatException;
import mulan.data.MultiLabelInstances;
import weka.core.Instance;
import weka.core.Instances;

/**
* Class implementing the SCUMBLE
*
* @author Jose Maria Moyano Murillo
*/
public class SCUMBLE extends MLDataMetric {

    /**
     * Constructor
     */
    public SCUMBLE() {
        super("SCUMBLE");
    }

    /**
     * Calculate metric value
     * 
     * @param mlData Multi-label dataset to which calculate the metric
     * @return Value of the metric
     */
    public double calculate(MultiLabelInstances mlData) {

        double SCUMBLE = 0.0;

        //        ImbalancedFeature [] imbalanced_data =  Utils.getImbalancedWithIR(mlData, Utils.getSortedByFrequency(Utils.getAppearancesPerLabel(mlData)));
        //        
        //        ImbalancedFeature[] new_imbalanced_data = new ImbalancedFeature[imbalanced_data.length];
        //        
        //        for(int i=0; i<imbalanced_data.length; i++){
        //            for(int j=0; j<imbalanced_data.length; j++){
        //                if(mlData.getLabelNames()[i].equals(imbalanced_data[j].getName())){
        //                    new_imbalanced_data[i] = imbalanced_data[j];
        //                }
        //            }
        //        }

        double[] ir = getIRperLabel(mlData);

        int nLabels = mlData.getNumLabels();
        Instances instances = mlData.getDataSet();
        double IRLmean = 0;
        int nActive = 0;
        double prod = 1;
        double sum = 0;

        int[] labelIndices = mlData.getLabelIndices();

        for (Instance inst : instances) {
            IRLmean = 0;
            prod = 1;
            nActive = 0;

            for (int l = 0; l < nLabels; l++) {
                if (inst.value(labelIndices[l]) == 1) {
                    prod *= ir[l];
                    IRLmean += ir[l];
                    nActive++;
                }
                //              else{
                //                 prod *= 0;
                //              }              
            }

            if (nActive == 0) {
                sum += 0;
            } else {
                IRLmean /= nActive;

                //               System.out.println(IRLmean);

                sum += 1 - (Math.pow(prod, 1.0 / nActive) / IRLmean);
            }

        }

        SCUMBLE = sum / mlData.getNumInstances();

        this.value = SCUMBLE;
        return value;
    }

    public double[] getIRperLabel(MultiLabelInstances mlData) {
        int[] appearances = getAppearances(mlData);

        //Get max appearance
        int max = appearances[0];
        for (int i = 1; i < appearances.length; i++) {
            if (appearances[i] > max) {
                max = appearances[i];
            }
        }

        //Calculate IR as maxFreq / freq[i]
        double[] ir = new double[mlData.getNumLabels()];
        for (int i = 0; i < mlData.getNumLabels(); i++) {
            ir[i] = (double) max / appearances[i];
        }

        return ir;
    }

    public int[] getAppearances(MultiLabelInstances mlData) {
        int[] appearances = new int[mlData.getNumLabels()];
        int[] labelIndices = mlData.getLabelIndices();

        for (Instance instance : mlData.getDataSet()) {
            for (int label = 0; label < mlData.getNumLabels(); label++) {
                if (instance.value(labelIndices[label]) == 1) {
                    appearances[label]++;
                }
            }
        }

        return appearances;
    }

    public static void main(String[] args) {
        try {
            MultiLabelInstances mlData = new MultiLabelInstances("Emotions.arff", "Emotions.xml");
            SCUMBLE scumble = new SCUMBLE();
            double value = scumble.calculate(mlData);
            System.out.println(value);

        } catch (InvalidDataFormatException e) {
            e.printStackTrace();
        }

    }

}