Examples with Clusters edu.ucsf.rbvi.clusterMaker2.internal.algorithms.attributeClusterers.Clusters used on opensource projects

Example 21 with Clusters

use of edu.ucsf.rbvi.clusterMaker2.internal.algorithms.attributeClusterers.Clusters in project clusterMaker2 by RBVI.

the class PAMTest method testSingletonCluster.

@Test
public void testSingletonCluster() {
    Double[] data = { .9, .9 };
    int k = 1;
    int[] ans = { 0 };
    CyMatrix mat = CyMatrixFactory.makeSmallMatrix(1, 2, data);
    PAM pam = new PAM(null, mat, DistanceMetric.CITYBLOCK);
    Clusters c = pam.cluster(k);
    assertEquals(c.getNumberOfClusters(), k);
}

Example 22 with Clusters

use of edu.ucsf.rbvi.clusterMaker2.internal.algorithms.attributeClusterers.Clusters in project clusterMaker2 by RBVI.

the class PAMTest method testLarge.

@Test
public void testLarge() {
    System.out.println("testLarge begin");
    Double[] data;
    int k = 8;
    int[] ans = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7 };
    BufferedReader reader = new BufferedReader(new InputStreamReader(PAMTest.class.getResourceAsStream("/pam_data.txt")));
    String line;
    ArrayList<Double> vectors = new ArrayList<Double>();
    int vectorWidth = 0;
    try {
        while ((line = reader.readLine()) != null) {
            String[] vector = line.split("\t");
            vectorWidth = vector.length;
            for (String v : vector) vectors.add(Double.parseDouble(v));
        }
    } catch (IOException e) {
        // TODO Auto-generated catch block
        e.printStackTrace();
    }
    data = new Double[vectors.size()];
    data = vectors.toArray(data);
    // BaseMatrix mat = new BaseMatrix(0, vectorWidth, data);
    CyMatrix mat = CyMatrixFactory.makeSmallMatrix(data.length / vectorWidth, vectorWidth, data);
    PAM pam = new PAM(null, mat, DistanceMetric.EUCLIDEAN);
    Clusters c = pam.cluster(k);
    System.out.println("testLarge end");
    assertEquals(c.getNumberOfClusters(), k);
    for (int i = 0; i < c.size(); ++i) {
        assertEquals(c.getClusterIndex(i), ans[i]);
    }
}

Example 23 with Clusters

use of edu.ucsf.rbvi.clusterMaker2.internal.algorithms.attributeClusterers.Clusters in project clusterMaker2 by RBVI.

the class MSplitSilhouetteCalculator method splitByAverageSilhouette.

public static Clusters splitByAverageSilhouette(Segregatable seg, int K, boolean forceSplit, Summarizer summarizer) {
    Clusters split = segregateByAverageSilhouette(seg, K, summarizer);
    if (!forceSplit) {
        // consider no split (k = 1)
        if (split.getCost() >= 1) {
            // cost >= 1  =>  average silhouette < 0  =>  no splitting is warranted
            split = seg.cluster(1);
            split.setCost(1.0);
        }
    }
    return split;
}

Example 24 with Clusters

use of edu.ucsf.rbvi.clusterMaker2.internal.algorithms.attributeClusterers.Clusters in project clusterMaker2 by RBVI.

the class MSplitSilhouetteCalculator method averageSilhouettes.

public static ArrayList<Double> averageSilhouettes(Subsegregatable sseg, Clusters clusters, int L, Summarizer summarizer) {
    int K = clusters.getNumberOfClusters();
    ArrayList<Double> splitSilhouettes = new ArrayList<Double>();
    int[][] partitions = clusters.getPartitions();
    // calculate the split silhouette of each cluster
    for (int kk = 0; kk < K; ++kk) {
        Clusters subclusters = segregateByAverageSilhouette(sseg.subset(partitions[kk]), L, summarizer);
        if (subclusters != null) {
            // cluster could be split further into subclusters
            splitSilhouettes.add(1 - subclusters.getCost());
        }
    }
    return splitSilhouettes;
}

Example 25 with Clusters

use of edu.ucsf.rbvi.clusterMaker2.internal.algorithms.attributeClusterers.Clusters in project clusterMaker2 by RBVI.

the class getClusters method findClusters.

public void findClusters(boolean general) {
    // used edges for DFS
    boolean[] used = new boolean[nodes.length];
    for (int i = 0; i < used.length; i++) used[i] = false;
    List<List<double[]>> clust = new ArrayList<List<double[]>>();
    List<List<String>> clustLabels = new ArrayList<List<String>>();
    List<List<Integer>> clustIDs = new ArrayList<List<Integer>>();
    List<List<Integer>> allNodes = new ArrayList<List<Integer>>();
    for (int j = 0; j < edges.length; j++) {
        if (edges[j][2] > maxDist)
            maxDist = edges[j][2];
    }
    int max = 0;
    // iterate across all rescaled SOM nodes
    for (int i = 0; i < nodes.length; i++) {
        // if node i has not been accessed, find all nodes connected to node i
        if (!used[i])
            connected(i);
        else
            // if node i has already been added to cluster, continue
            continue;
        // populate cluster arrays
        List<double[]> allIndices = new ArrayList<double[]>();
        List<String> allLabels = new ArrayList<String>();
        List<Integer> allIDs = new ArrayList<Integer>();
        List<Integer> indices = new ArrayList<Integer>();
        for (int index : currClust) {
            if (!general) {
                // if TR clusters
                double[] info = new double[nodes[index].getPoint().length];
                for (int k = 0; k < info.length; k++) info[k] = nodes[index].getPoint()[k];
                allIndices.add(info);
            } else {
                // If general numerical clusters
                for (int h = 0; h < ids[index].size(); h++) {
                    double[] info = new double[nodes[index].getPoint().length + 1];
                    /* String lab = new StringTokenizer(labels[index].get(h).toString(),",").nextToken();
                                try{
                                    int ID = Integer.parseInt(lab);
                                    info[0] = ID;
                                }catch(NumberFormatException err) {info[0]=1;}   */
                    // indices.add(index);
                    info[0] = index;
                    for (int k = 1; k < info.length; k++) info[k] = nodes[index].getPoint()[k - 1];
                    int id = ids[index].get(h).intValue();
                    String[] tokens = s.input[id].getDesc().split(",");
                    if (s.benchmark) {
                        int label = Integer.valueOf(tokens[0]);
                        // if clusters are known (benchmarking), how many are there?
                        if (label > max)
                            max = (int) label;
                    }
                    allIndices.add(info);
                    // allLabels.add(labels[index].get(h).toString());
                    allIDs.add(ids[index].get(h));
                }
            }
            // node 'index' is now used
            used[index] = true;
        }
        allNodes.add(indices);
        // add node indices to current cluster
        clust.add(allIndices);
        // add labels to current cluster
        clustLabels.add(allLabels);
        clustIDs.add(allIDs);
        currClust.clear();
    }
    clusters = new cluster[clust.size()];
    for (int i = 0; i < clusters.length; i++) {
        clusters[i] = new cluster(clust.get(i), clustLabels.get(i), clustIDs.get(i), allNodes.get(i));
        if (s.benchmark) {
            // if benchmarking, store original clusters
            origClusters = (List<Integer>[]) Array.newInstance(List.class, max);
            if (general) {
                for (int j = 0; j < clusters[i].labels.size(); j++) {
                    String[] tokens = clusters[i].labels.get(j).toString().split(",");
                    int label = Integer.valueOf(tokens[0]);
                    if (origClusters[label - 1] == null)
                        origClusters[label - 1] = new ArrayList<Integer>();
                    origClusters[label - 1].add(label);
                }
            }
        }
    }
    currClust.clear();
}