Example usage for weka.core MinkowskiDistance MinkowskiDistance

Introduction

In this page you can find the example usage for weka.core MinkowskiDistance MinkowskiDistance.

Prototype

public MinkowskiDistance(Instances data) 

Source Link

Document

Constructs an Minkowski Distance object and automatically initializes the ranges.

Usage

From source file:lu.lippmann.cdb.lab.mds.ClassicMDS.java

License:Open Source License

/**
 * //from w  w  w .  j  a  v  a  2s .  c  o  m
 */
public static CollapsedInstances distanceBetweenInstances(final Instances instances,
        final MDSDistancesEnum distEnum, final int maxInstances, final boolean ignoreClassInDistance)
        throws Exception {
    KmeansResult mapCentroids = null;

    final NormalizableDistance usedDist;
    if (distEnum.equals(MDSDistancesEnum.EUCLIDEAN)) {
        usedDist = new EuclideanDistance(instances);
        //usedDist.setDontNormalize(true);
        //usedDist.setAttributeIndices("1");
        //usedDist.setInvertSelection(true);
    } else if (distEnum.equals(MDSDistancesEnum.MANHATTAN))
        usedDist = new ManhattanDistance(instances);
    else if (distEnum.equals(MDSDistancesEnum.MINKOWSKI)) {
        usedDist = new MinkowskiDistance(instances);
        final String[] parameters = MDSDistancesEnum.MINKOWSKI.getParameters();
        //Change order
        double order = Double.valueOf(parameters[0]).doubleValue();
        ((MinkowskiDistance) usedDist).setOrder(order);
    } else if (distEnum.equals(MDSDistancesEnum.CHEBYSHEV))
        usedDist = new ChebyshevDistance(instances);
    //else if (distEnum.equals(MDSDistancesEnum.DT)) usedDist=new DTDistance(instances);
    else
        throw new IllegalStateException();

    final int numInstances = instances.numInstances();
    final boolean collapsed = (numInstances > maxInstances)
            && (distEnum.equals(MDSDistancesEnum.EUCLIDEAN) || distEnum.equals(MDSDistancesEnum.MANHATTAN));

    SimpleMatrix distances;

    //Ignore class in distance
    if (ignoreClassInDistance && instances.classIndex() != -1) {
        usedDist.setAttributeIndices("" + (instances.classIndex() + 1));
        usedDist.setInvertSelection(true);
    }

    int numCollapsedInstances = numInstances;
    if (collapsed) {
        //Compute distance with centroids using K-means with K=MAX_INSTANCES
        mapCentroids = getSimplifiedInstances(instances, usedDist, maxInstances);

        final List<Instance> centroids = mapCentroids.getCentroids();
        numCollapsedInstances = centroids.size();

        distances = new SimpleMatrix(numCollapsedInstances, numCollapsedInstances);

        for (int i = 0; i < numCollapsedInstances; i++) {
            for (int j = i + 1; j < numCollapsedInstances; j++) {
                double dist = usedDist.distance(centroids.get(i), centroids.get(j));
                distances.set(i, j, dist);
                distances.set(j, i, dist);
            }
        }
    } else {
        distances = new SimpleMatrix(numCollapsedInstances, numCollapsedInstances);
        for (int i = 0; i < numCollapsedInstances; i++) {
            for (int j = i + 1; j < numCollapsedInstances; j++) {
                double dist = usedDist.distance(instances.get(i), instances.get(j));
                distances.set(i, j, dist);
                distances.set(j, i, dist);
            }
        }
    }
    return new CollapsedInstances(instances, mapCentroids, distances, collapsed);
}

---

From source file:nl.uva.sne.classifiers.Hierarchical.java

@Override
public Map<String, String> cluster(String inDir) throws IOException, ParseException {
    try {/* ww w .j ava2  s . c o  m*/

        Instances data = ClusterUtils.terms2Instances(inDir, false);

        //            ArffSaver s = new ArffSaver();
        //            s.setInstances(data);
        //            s.setFile(new File(inDir+"/dataset.arff"));
        //            s.writeBatch();

        DistanceFunction df;
        //            SimpleKMeans currently only supports the Euclidean and Manhattan distances.
        switch (distanceFunction) {
        case "Minkowski":
            df = new MinkowskiDistance(data);
            break;
        case "Euclidean":
            df = new EuclideanDistance(data);
            break;
        case "Chebyshev":
            df = new ChebyshevDistance(data);
            break;
        case "Manhattan":
            df = new ManhattanDistance(data);
            break;
        default:
            df = new EuclideanDistance(data);
            break;
        }

        Logger.getLogger(Hierarchical.class.getName()).log(Level.INFO, "Start clusteing");

        weka.clusterers.HierarchicalClusterer clusterer = new HierarchicalClusterer();
        clusterer.setOptions(new String[] { "-L", "COMPLETE" });
        clusterer.setDebug(true);
        clusterer.setNumClusters(numOfClusters);
        clusterer.setDistanceFunction(df);
        clusterer.setDistanceIsBranchLength(true);
        clusterer.setPrintNewick(false);

        weka.clusterers.FilteredClusterer fc = new weka.clusterers.FilteredClusterer();
        String[] options = new String[2];
        options[0] = "-R"; // "range"
        options[1] = "1"; // we want to ignore the attribute that is in the position '1'
        Remove remove = new Remove(); // new instance of filter
        remove.setOptions(options); // set options

        fc.setFilter(remove); //add filter to remove attributes
        fc.setClusterer(clusterer); //bind FilteredClusterer to original clusterer
        fc.buildClusterer(data);

        //             // Print normal
        //        clusterer.setPrintNewick(false);
        //        System.out.println(clusterer.graph());
        //        // Print Newick
        //        clusterer.setPrintNewick(true);
        //        System.out.println(clusterer.graph());
        //
        //        // Let's try to show this clustered data!
        //        JFrame mainFrame = new JFrame("Weka Test");
        //        mainFrame.setSize(600, 400);
        //        mainFrame.setDefaultCloseOperation(JFrame.EXIT_ON_CLOSE);
        //        Container content = mainFrame.getContentPane();
        //        content.setLayout(new GridLayout(1, 1));
        //
        //        HierarchyVisualizer visualizer = new HierarchyVisualizer(clusterer.graph());
        //        content.add(visualizer);
        //
        //        mainFrame.setVisible(true);
        return ClusterUtils.bulidClusters(clusterer, data, inDir);

    } catch (Exception ex) {
        Logger.getLogger(Hierarchical.class.getName()).log(Level.SEVERE, null, ex);
    }
    return null;
}

---