lu.lippmann.cdb.ext.hydviga.gaps.GapFillerRegressions.java Source code

Introduction

Here is the source code for lu.lippmann.cdb.ext.hydviga.gaps.GapFillerRegressions.java

Source

/**
 * Copyright 2014-2016 LIST (Luxembourg Institute of Science and Technology), all right reserved.
 * Authorship : Olivier PARISOT, Yoanne DIDRY
 * Licensed under GNU General Public License version 3
 */
package lu.lippmann.cdb.ext.hydviga.gaps;

import lu.lippmann.cdb.lab.regression.Regression;
import lu.lippmann.cdb.weka.*;
import weka.core.Instances;

/**
 * TimeSeriesGapFillerRegressions.
 * 
 * @author Olivier PARISOT
 *
 */
public class GapFillerRegressions extends GapFiller {
    //
    // Constructors
    //

    /**
     * Constructor.    
     */
    GapFillerRegressions(final boolean wdt) {
        super(wdt);
    }

    //
    // Instance methods
    //

    /**
     * {@inheritDoc}
     */
    @Override
    Instances fillGaps0(final Instances ds) throws Exception {
        final Instances newds = WekaDataProcessingUtil.buildDataSetWithoutConstantAttributes(ds);
        final int numInstances = newds.numInstances();

        final int attrWithMissingIdx = WekaDataStatsUtil.getFirstAttributeWithMissingValue(newds);
        if (attrWithMissingIdx == -1)
            throw new IllegalStateException();

        final Instances trainingSet = new Instances(newds, 0);
        for (int i = 0; i < numInstances; i++) {
            if (!newds.instance(i).hasMissingValue())
                trainingSet.add(newds.instance(i));
        }
        //System.out.println(trainingSet);

        final Regression reg = new Regression(trainingSet, attrWithMissingIdx);
        final double[] coeffs = reg.getCoe();
        //System.out.println(reg.getR2());
        //System.out.println(reg.getCoeDesc());

        for (int i = 0; i < numInstances; i++) {
            if (newds.instance(i).isMissing(attrWithMissingIdx)) {
                double newval = coeffs[0];
                for (int j = 1; j < trainingSet.numAttributes(); j++) {
                    if (j == attrWithMissingIdx)
                        continue;

                    final String attrName = trainingSet.attribute(j).name();

                    //System.out.println(reg.getCoef(attrName)+" * "+attrName);

                    newval += reg.getCoef(attrName) * newds.instance(i).value(newds.attribute(attrName));
                }
                //System.out.println("oldval -> "+newds.instance(i).value(attrWithMissingIdx));
                //System.out.println("newval -> "+newval);
                newds.instance(i).setValue(attrWithMissingIdx, newval);
            }
        }

        //System.out.println("corrected -> "+newds);

        this.model = reg.getCoeDesc();

        return newds;
    }

    /**
     * {@inheritDoc}
     */
    @Override
    public boolean hasExplicitModel() {
        return true;
    }

}