Examples with Clustering de.lmu.ifi.dbs.elki.data.Clustering used on opensource projects

Example 46 with Clustering

use of de.lmu.ifi.dbs.elki.data.Clustering in project elki by elki-project.

the class KMeansMacQueen method run.

@Override
public Clustering<KMeansModel> run(Database database, Relation<V> relation) {
    if (relation.size() <= 0) {
        return new Clustering<>("k-Means Clustering", "kmeans-clustering");
    }
    // Choose initial means
    if (LOG.isStatistics()) {
        LOG.statistics(new StringStatistic(KEY + ".initialization", initializer.toString()));
    }
    double[][] means = initializer.chooseInitialMeans(database, relation, k, getDistanceFunction());
    List<ModifiableDBIDs> clusters = new ArrayList<>();
    for (int i = 0; i < k; i++) {
        clusters.add(DBIDUtil.newHashSet((int) (relation.size() * 2. / k)));
    }
    WritableIntegerDataStore assignment = DataStoreUtil.makeIntegerStorage(relation.getDBIDs(), DataStoreFactory.HINT_TEMP | DataStoreFactory.HINT_HOT, -1);
    double[] varsum = new double[k];
    IndefiniteProgress prog = LOG.isVerbose() ? new IndefiniteProgress("K-Means iteration", LOG) : null;
    DoubleStatistic varstat = LOG.isStatistics() ? new DoubleStatistic(this.getClass().getName() + ".variance-sum") : null;
    // Iterate MacQueen
    int iteration = 0;
    for (; maxiter <= 0 || iteration < maxiter; iteration++) {
        LOG.incrementProcessed(prog);
        boolean changed = macQueenIterate(relation, means, clusters, assignment, varsum);
        logVarstat(varstat, varsum);
        if (!changed) {
            break;
        }
    }
    LOG.setCompleted(prog);
    if (LOG.isStatistics()) {
        LOG.statistics(new LongStatistic(KEY + ".iterations", iteration));
    }
    Clustering<KMeansModel> result = new Clustering<>("k-Means Clustering", "kmeans-clustering");
    for (int i = 0; i < clusters.size(); i++) {
        DBIDs ids = clusters.get(i);
        if (ids.size() == 0) {
            continue;
        }
        KMeansModel model = new KMeansModel(means[i], varsum[i]);
        result.addToplevelCluster(new Cluster<>(ids, model));
    }
    return result;
}

Example 47 with Clustering

use of de.lmu.ifi.dbs.elki.data.Clustering in project elki by elki-project.

the class KMeansSort method run.

@Override
public Clustering<KMeansModel> run(Database database, Relation<V> relation) {
    if (relation.size() <= 0) {
        return new Clustering<>("k-Means Clustering", "kmeans-clustering");
    }
    // Choose initial means
    if (LOG.isStatistics()) {
        LOG.statistics(new StringStatistic(KEY + ".initialization", initializer.toString()));
    }
    double[][] means = initializer.chooseInitialMeans(database, relation, k, getDistanceFunction());
    // Setup cluster assignment store
    List<ModifiableDBIDs> clusters = new ArrayList<>();
    for (int i = 0; i < k; i++) {
        clusters.add(DBIDUtil.newHashSet((int) (relation.size() * 2. / k)));
    }
    WritableIntegerDataStore assignment = DataStoreUtil.makeIntegerStorage(relation.getDBIDs(), DataStoreFactory.HINT_TEMP | DataStoreFactory.HINT_HOT, -1);
    double[] varsum = new double[k];
    // Cluster distances
    double[][] cdist = new double[k][k];
    int[][] cnum = new int[k][k - 1];
    IndefiniteProgress prog = LOG.isVerbose() ? new IndefiniteProgress("K-Means iteration", LOG) : null;
    DoubleStatistic varstat = LOG.isStatistics() ? new DoubleStatistic(this.getClass().getName() + ".variance-sum") : null;
    LongStatistic diststat = LOG.isStatistics() ? new LongStatistic(KEY + ".distance-computations") : null;
    int iteration = 0;
    for (; maxiter <= 0 || iteration < maxiter; iteration++) {
        LOG.incrementProcessed(prog);
        recomputeSeperation(means, cdist, cnum, diststat);
        boolean changed = assignToNearestCluster(relation, means, clusters, assignment, varsum, cdist, cnum, diststat);
        logVarstat(varstat, varsum);
        if (LOG.isStatistics()) {
            LOG.statistics(diststat);
        }
        // Stop if no cluster assignment changed.
        if (!changed) {
            break;
        }
        // Recompute means.
        means = means(clusters, means, relation);
    }
    LOG.setCompleted(prog);
    if (LOG.isStatistics()) {
        LOG.statistics(new LongStatistic(KEY + ".iterations", iteration));
    }
    // Wrap result
    Clustering<KMeansModel> result = new Clustering<>("k-Means Clustering", "kmeans-clustering");
    for (int i = 0; i < clusters.size(); i++) {
        DBIDs ids = clusters.get(i);
        if (ids.size() == 0) {
            continue;
        }
        KMeansModel model = new KMeansModel(means[i], varsum[i]);
        result.addToplevelCluster(new Cluster<>(ids, model));
    }
    return result;
}

Example 48 with Clustering

use of de.lmu.ifi.dbs.elki.data.Clustering in project elki by elki-project.

the class LSDBC method run.

/**
 * Run the LSDBC algorithm
 *
 * @param database Database to process
 * @param relation Data relation
 * @return Clustering result
 */
public Clustering<Model> run(Database database, Relation<O> relation) {
    StepProgress stepprog = LOG.isVerbose() ? new StepProgress("LSDBC", 3) : null;
    final int dim = RelationUtil.dimensionality(relation);
    final double factor = FastMath.pow(2., alpha / dim);
    final DBIDs ids = relation.getDBIDs();
    LOG.beginStep(stepprog, 1, "Materializing kNN neighborhoods");
    KNNQuery<O> knnq = DatabaseUtil.precomputedKNNQuery(database, relation, getDistanceFunction(), k);
    LOG.beginStep(stepprog, 2, "Sorting by density");
    WritableDoubleDataStore dens = DataStoreUtil.makeDoubleStorage(ids, DataStoreFactory.HINT_HOT | DataStoreFactory.HINT_TEMP);
    fillDensities(knnq, ids, dens);
    ArrayModifiableDBIDs sids = DBIDUtil.newArray(ids);
    sids.sort(new DataStoreUtil.AscendingByDoubleDataStore(dens));
    LOG.beginStep(stepprog, 3, "Computing clusters");
    // Setup progress logging
    final FiniteProgress progress = LOG.isVerbose() ? new FiniteProgress("LSDBC Clustering", ids.size(), LOG) : null;
    final IndefiniteProgress clusprogress = LOG.isVerbose() ? new IndefiniteProgress("Number of clusters found", LOG) : null;
    // (Temporary) store the cluster ID assigned.
    final WritableIntegerDataStore clusterids = DataStoreUtil.makeIntegerStorage(ids, DataStoreFactory.HINT_TEMP, UNPROCESSED);
    // Note: these are not exact, as objects may be stolen from noise.
    final IntArrayList clustersizes = new IntArrayList();
    // Unprocessed dummy value.
    clustersizes.add(0);
    // Noise counter.
    clustersizes.add(0);
    // Implementation Note: using Integer objects should result in
    // reduced memory use in the HashMap!
    int clusterid = NOISE + 1;
    // Iterate over all objects in the database.
    for (DBIDIter id = sids.iter(); id.valid(); id.advance()) {
        // Skip already processed ids.
        if (clusterids.intValue(id) != UNPROCESSED) {
            continue;
        }
        // Evaluate Neighborhood predicate
        final KNNList neighbors = knnq.getKNNForDBID(id, k);
        // Evaluate Core-Point predicate:
        if (isLocalMaximum(neighbors.getKNNDistance(), neighbors, dens)) {
            double mindens = factor * neighbors.getKNNDistance();
            clusterids.putInt(id, clusterid);
            clustersizes.add(expandCluster(clusterid, clusterids, knnq, neighbors, mindens, progress));
            // start next cluster on next iteration.
            ++clusterid;
            if (clusprogress != null) {
                clusprogress.setProcessed(clusterid, LOG);
            }
        } else {
            // otherwise, it's a noise point
            clusterids.putInt(id, NOISE);
            clustersizes.set(NOISE, clustersizes.getInt(NOISE) + 1);
        }
        // We've completed this element
        LOG.incrementProcessed(progress);
    }
    // Finish progress logging.
    LOG.ensureCompleted(progress);
    LOG.setCompleted(clusprogress);
    LOG.setCompleted(stepprog);
    // Transform cluster ID mapping into a clustering result:
    ArrayList<ArrayModifiableDBIDs> clusterlists = new ArrayList<>(clusterid);
    // add storage containers for clusters
    for (int i = 0; i < clustersizes.size(); i++) {
        clusterlists.add(DBIDUtil.newArray(clustersizes.getInt(i)));
    }
    // do the actual inversion
    for (DBIDIter id = ids.iter(); id.valid(); id.advance()) {
        // Negative values are non-core points:
        int cid = clusterids.intValue(id);
        int cluster = Math.abs(cid);
        clusterlists.get(cluster).add(id);
    }
    clusterids.destroy();
    Clustering<Model> result = new Clustering<>("LSDBC", "lsdbc-clustering");
    for (int cid = NOISE; cid < clusterlists.size(); cid++) {
        boolean isNoise = (cid == NOISE);
        Cluster<Model> c;
        c = new Cluster<Model>(clusterlists.get(cid), isNoise, ClusterModel.CLUSTER);
        result.addToplevelCluster(c);
    }
    return result;
}

Example 49 with Clustering

use of de.lmu.ifi.dbs.elki.data.Clustering in project elki by elki-project.

the class ORCLUS method run.

/**
 * Performs the ORCLUS algorithm on the given database.
 *
 * @param database Database
 * @param relation Relation
 */
public Clustering<Model> run(Database database, Relation<V> relation) {
    // current dimensionality associated with each seed
    int dim_c = RelationUtil.dimensionality(relation);
    if (dim_c < l) {
        throw new IllegalStateException("Dimensionality of data < parameter l! " + "(" + dim_c + " < " + l + ")");
    }
    // current number of seeds
    int k_c = Math.min(relation.size(), k_i * k);
    // pick k0 > k points from the database
    List<ORCLUSCluster> clusters = initialSeeds(relation, k_c);
    double beta = FastMath.exp(-FastMath.log(dim_c / (double) l) * FastMath.log(1 / alpha) / FastMath.log(k_c / (double) k));
    IndefiniteProgress cprogress = LOG.isVerbose() ? new IndefiniteProgress("Current number of clusters:", LOG) : null;
    while (k_c > k) {
        // find partitioning induced by the seeds of the clusters
        assign(relation, clusters);
        // determine current subspace associated with each cluster
        for (ORCLUSCluster cluster : clusters) {
            if (cluster.objectIDs.size() > 0) {
                cluster.basis = findBasis(relation, cluster, dim_c);
            }
        }
        // reduce number of seeds and dimensionality associated with
        // each seed
        k_c = (int) Math.max(k, k_c * alpha);
        dim_c = (int) Math.max(l, dim_c * beta);
        merge(relation, clusters, k_c, dim_c, cprogress);
        if (cprogress != null) {
            cprogress.setProcessed(clusters.size(), LOG);
        }
    }
    assign(relation, clusters);
    LOG.setCompleted(cprogress);
    // get the result
    Clustering<Model> r = new Clustering<>("ORCLUS clustering", "orclus-clustering");
    for (ORCLUSCluster c : clusters) {
        r.addToplevelCluster(new Cluster<Model>(c.objectIDs, ClusterModel.CLUSTER));
    }
    return r;
}

Example 50 with Clustering

use of de.lmu.ifi.dbs.elki.data.Clustering in project elki by elki-project.

the class CLARA method run.

@Override
public Clustering<MedoidModel> run(Database database, Relation<V> relation) {
    if (relation.size() <= 0) {
        return new Clustering<>("CLARA Clustering", "clara-clustering");
    }
    DBIDs ids = relation.getDBIDs();
    DistanceQuery<V> distQ = database.getDistanceQuery(relation, getDistanceFunction());
    double best = Double.POSITIVE_INFINITY;
    ArrayModifiableDBIDs bestmedoids = null;
    WritableIntegerDataStore bestclusters = null;
    Random rnd = random.getSingleThreadedRandom();
    FiniteProgress prog = LOG.isVerbose() ? new FiniteProgress("Processing random samples", numsamples, LOG) : null;
    for (int j = 0; j < numsamples; j++) {
        DBIDs rids = DBIDUtil.randomSample(ids, sampling, rnd);
        // FIXME: precompute and use a distance matrix for this sample!
        // Choose initial medoids
        ArrayModifiableDBIDs medoids = DBIDUtil.newArray(initializer.chooseInitialMedoids(k, rids, distQ));
        // Setup cluster assignment store
        WritableIntegerDataStore assignment = DataStoreUtil.makeIntegerStorage(ids, DataStoreFactory.HINT_HOT | DataStoreFactory.HINT_TEMP, -1);
        new /* PAM */
        Instance(distQ, rids, assignment).run(medoids, maxiter);
        double score = assignRemainingToNearestCluster(medoids, ids, rids, assignment, distQ);
        if (score < best) {
            best = score;
            bestmedoids = medoids;
            bestclusters = assignment;
        }
        LOG.incrementProcessed(prog);
    }
    LOG.ensureCompleted(prog);
    ArrayModifiableDBIDs[] clusters = ClusteringAlgorithmUtil.partitionsFromIntegerLabels(ids, bestclusters, k);
    // Wrap result
    Clustering<MedoidModel> result = new Clustering<>("CLARA Clustering", "clara-clustering");
    for (DBIDArrayIter it = bestmedoids.iter(); it.valid(); it.advance()) {
        MedoidModel model = new MedoidModel(DBIDUtil.deref(it));
        result.addToplevelCluster(new Cluster<>(clusters[it.getOffset()], model));
    }
    return result;
}