Examples with LongStatistic de.lmu.ifi.dbs.elki.logging.statistics.LongStatistic used on opensource projects

Example 31 with LongStatistic

use of de.lmu.ifi.dbs.elki.logging.statistics.LongStatistic in project elki by elki-project.

the class EvaluatePBMIndex method evaluateClustering.

/**
 * Evaluate a single clustering.
 *
 * @param db Database
 * @param rel Data relation
 * @param c Clustering
 * @return PBM
 */
public double evaluateClustering(Database db, Relation<? extends NumberVector> rel, Clustering<?> c) {
    List<? extends Cluster<?>> clusters = c.getAllClusters();
    NumberVector[] centroids = new NumberVector[clusters.size()];
    int ignorednoise = EvaluateSimplifiedSilhouette.centroids(rel, clusters, centroids, noiseHandling);
    // Build global centroid and cluster count:
    final int dim = RelationUtil.dimensionality(rel);
    Centroid overallCentroid = new Centroid(dim);
    EvaluateVarianceRatioCriteria.globalCentroid(overallCentroid, rel, clusters, centroids, noiseHandling);
    // Maximum distance between centroids:
    double max = 0;
    for (int i = 0; i < centroids.length; i++) {
        if (centroids[i] == null && noiseHandling != NoiseHandling.TREAT_NOISE_AS_SINGLETONS) {
            continue;
        }
        for (int j = i + 1; j < centroids.length; j++) {
            if (centroids[j] == null && noiseHandling != NoiseHandling.TREAT_NOISE_AS_SINGLETONS) {
                continue;
            }
            if (centroids[i] == null && centroids[j] == null) {
                // Need to compute pairwise distances of noise clusters.
                for (DBIDIter iti = clusters.get(i).getIDs().iter(); iti.valid(); iti.advance()) {
                    for (DBIDIter itj = clusters.get(j).getIDs().iter(); itj.valid(); itj.advance()) {
                        double dist = distanceFunction.distance(rel.get(iti), rel.get(itj));
                        max = dist > max ? dist : max;
                    }
                }
            } else if (centroids[i] == null) {
                for (DBIDIter iti = clusters.get(i).getIDs().iter(); iti.valid(); iti.advance()) {
                    double dist = distanceFunction.distance(rel.get(iti), centroids[j]);
                    max = dist > max ? dist : max;
                }
            } else if (centroids[j] == null) {
                for (DBIDIter itj = clusters.get(j).getIDs().iter(); itj.valid(); itj.advance()) {
                    double dist = distanceFunction.distance(centroids[i], rel.get(itj));
                    max = dist > max ? dist : max;
                }
            } else {
                double dist = distanceFunction.distance(centroids[i], centroids[j]);
                max = dist > max ? dist : max;
            }
        }
    }
    // a: Distance to own centroid
    // b: Distance to overall centroid
    double a = 0, b = 0;
    Iterator<? extends Cluster<?>> ci = clusters.iterator();
    for (int i = 0; ci.hasNext(); i++) {
        Cluster<?> cluster = ci.next();
        if (cluster.size() <= 1 || cluster.isNoise()) {
            switch(noiseHandling) {
                case IGNORE_NOISE:
                    // Ignored
                    continue;
                case TREAT_NOISE_AS_SINGLETONS:
                    // Singletons: a = 0 by definition.
                    for (DBIDIter it = cluster.getIDs().iter(); it.valid(); it.advance()) {
                        b += SquaredEuclideanDistanceFunction.STATIC.distance(overallCentroid, rel.get(it));
                    }
                    // with NEXT cluster.
                    continue;
                case MERGE_NOISE:
                    // Treat like a cluster below:
                    break;
            }
        }
        for (DBIDIter it = cluster.getIDs().iter(); it.valid(); it.advance()) {
            NumberVector obj = rel.get(it);
            a += distanceFunction.distance(centroids[i], obj);
            b += distanceFunction.distance(overallCentroid, obj);
        }
    }
    final double pbm = FastMath.pow((1. / centroids.length) * (b / a) * max, 2.);
    if (LOG.isStatistics()) {
        LOG.statistics(new StringStatistic(key + ".pbm.noise-handling", noiseHandling.toString()));
        if (ignorednoise > 0) {
            LOG.statistics(new LongStatistic(key + ".pbm.ignored", ignorednoise));
        }
        LOG.statistics(new DoubleStatistic(key + ".pbm", pbm));
    }
    EvaluationResult ev = EvaluationResult.findOrCreate(db.getHierarchy(), c, "Internal Clustering Evaluation", "internal evaluation");
    MeasurementGroup g = ev.findOrCreateGroup("Distance-based Evaluation");
    g.addMeasure("PBM-Index", pbm, 0., Double.POSITIVE_INFINITY, 0., false);
    db.getHierarchy().resultChanged(ev);
    return pbm;
}

Example 32 with LongStatistic

use of de.lmu.ifi.dbs.elki.logging.statistics.LongStatistic in project elki by elki-project.

the class EvaluateSilhouette method evaluateClustering.

/**
 * Evaluate a single clustering.
 *
 * @param db Database
 * @param rel Data relation
 * @param dq Distance query
 * @param c Clustering
 * @return Average silhouette
 */
public double evaluateClustering(Database db, Relation<O> rel, DistanceQuery<O> dq, Clustering<?> c) {
    List<? extends Cluster<?>> clusters = c.getAllClusters();
    MeanVariance msil = new MeanVariance();
    int ignorednoise = 0;
    for (Cluster<?> cluster : clusters) {
        // Note: we treat 1-element clusters the same as noise.
        if (cluster.size() <= 1 || cluster.isNoise()) {
            switch(noiseOption) {
                case IGNORE_NOISE:
                    ignorednoise += cluster.size();
                    // Ignore noise elements
                    continue;
                case TREAT_NOISE_AS_SINGLETONS:
                    // As suggested in Rousseeuw, we use 0 for singletons.
                    msil.put(0., cluster.size());
                    continue;
                case MERGE_NOISE:
                    // Treat as cluster below
                    break;
            }
        }
        ArrayDBIDs ids = DBIDUtil.ensureArray(cluster.getIDs());
        // temporary storage.
        double[] as = new double[ids.size()];
        DBIDArrayIter it1 = ids.iter(), it2 = ids.iter();
        for (it1.seek(0); it1.valid(); it1.advance()) {
            // a: In-cluster distances
            // Already computed distances
            double a = as[it1.getOffset()];
            for (it2.seek(it1.getOffset() + 1); it2.valid(); it2.advance()) {
                final double dist = dq.distance(it1, it2);
                a += dist;
                as[it2.getOffset()] += dist;
            }
            a /= (ids.size() - 1);
            // b: minimum average distance to other clusters:
            double b = Double.POSITIVE_INFINITY;
            for (Cluster<?> ocluster : clusters) {
                if (ocluster == /* yes, reference identity */
                cluster) {
                    // Same cluster
                    continue;
                }
                if (ocluster.size() <= 1 || ocluster.isNoise()) {
                    switch(noiseOption) {
                        case IGNORE_NOISE:
                            // Ignore noise elements
                            continue;
                        case TREAT_NOISE_AS_SINGLETONS:
                            // Treat noise cluster as singletons:
                            for (DBIDIter it3 = ocluster.getIDs().iter(); it3.valid(); it3.advance()) {
                                final double dist = dq.distance(it1, it3);
                                // Minimum average
                                b = dist < b ? dist : b;
                            }
                            continue;
                        case MERGE_NOISE:
                            // Treat as cluster below
                            break;
                    }
                }
                final DBIDs oids = ocluster.getIDs();
                double btmp = 0.;
                for (DBIDIter it3 = oids.iter(); it3.valid(); it3.advance()) {
                    btmp += dq.distance(it1, it3);
                }
                // Average
                btmp /= oids.size();
                // Minimum average
                b = btmp < b ? btmp : b;
            }
            // One cluster only?
            b = b < Double.POSITIVE_INFINITY ? b : a;
            msil.put((b - a) / (b > a ? b : a));
        }
    }
    double penalty = 1.;
    // Only if {@link NoiseHandling#IGNORE_NOISE}:
    if (penalize && ignorednoise > 0) {
        penalty = (rel.size() - ignorednoise) / (double) rel.size();
    }
    final double meansil = penalty * msil.getMean();
    final double stdsil = penalty * msil.getSampleStddev();
    if (LOG.isStatistics()) {
        LOG.statistics(new StringStatistic(key + ".silhouette.noise-handling", noiseOption.toString()));
        if (ignorednoise > 0) {
            LOG.statistics(new LongStatistic(key + ".silhouette.noise", ignorednoise));
        }
        LOG.statistics(new DoubleStatistic(key + ".silhouette.mean", meansil));
        LOG.statistics(new DoubleStatistic(key + ".silhouette.stddev", stdsil));
    }
    EvaluationResult ev = EvaluationResult.findOrCreate(db.getHierarchy(), c, "Internal Clustering Evaluation", "internal evaluation");
    MeasurementGroup g = ev.findOrCreateGroup("Distance-based Evaluation");
    g.addMeasure("Silhouette +-" + FormatUtil.NF2.format(stdsil), meansil, -1., 1., 0., false);
    db.getHierarchy().resultChanged(ev);
    return meansil;
}

Example 33 with LongStatistic

use of de.lmu.ifi.dbs.elki.logging.statistics.LongStatistic in project elki by elki-project.

the class EvaluateSimplifiedSilhouette method evaluateClustering.

/**
 * Evaluate a single clustering.
 *
 * @param db Database
 * @param rel Data relation
 * @param c Clustering
 * @return Mean simplified silhouette
 */
public double evaluateClustering(Database db, Relation<? extends NumberVector> rel, Clustering<?> c) {
    List<? extends Cluster<?>> clusters = c.getAllClusters();
    NumberVector[] centroids = new NumberVector[clusters.size()];
    int ignorednoise = centroids(rel, clusters, centroids, noiseOption);
    MeanVariance mssil = new MeanVariance();
    Iterator<? extends Cluster<?>> ci = clusters.iterator();
    for (int i = 0; ci.hasNext(); i++) {
        Cluster<?> cluster = ci.next();
        if (cluster.size() <= 1) {
            // As suggested in Rousseeuw, we use 0 for singletons.
            mssil.put(0., cluster.size());
            continue;
        }
        if (cluster.isNoise()) {
            switch(noiseOption) {
                case IGNORE_NOISE:
                    // Ignore elements
                    continue;
                case TREAT_NOISE_AS_SINGLETONS:
                    // As suggested in Rousseeuw, we use 0 for singletons.
                    mssil.put(0., cluster.size());
                    continue;
                case MERGE_NOISE:
                    // Treat as cluster below
                    break;
            }
        }
        // Cluster center:
        final NumberVector center = centroids[i];
        assert (center != null);
        for (DBIDIter it = cluster.getIDs().iter(); it.valid(); it.advance()) {
            NumberVector obj = rel.get(it);
            // a: Distance to own centroid
            double a = distance.distance(center, obj);
            // b: Distance to other clusters centroids:
            double min = Double.POSITIVE_INFINITY;
            Iterator<? extends Cluster<?>> cj = clusters.iterator();
            for (int j = 0; cj.hasNext(); j++) {
                Cluster<?> ocluster = cj.next();
                if (i == j) {
                    continue;
                }
                NumberVector other = centroids[j];
                if (other == null) {
                    // Noise!
                    switch(noiseOption) {
                        case IGNORE_NOISE:
                            continue;
                        case TREAT_NOISE_AS_SINGLETONS:
                            // Treat each object like a centroid!
                            for (DBIDIter it2 = ocluster.getIDs().iter(); it2.valid(); it2.advance()) {
                                double dist = distance.distance(rel.get(it2), obj);
                                min = dist < min ? dist : min;
                            }
                            continue;
                        case MERGE_NOISE:
                            // Treat as cluster below, but should not be reachable.
                            break;
                    }
                }
                // Clusters: use centroid.
                double dist = distance.distance(other, obj);
                min = dist < min ? dist : min;
            }
            // One 'real' cluster only?
            min = min < Double.POSITIVE_INFINITY ? min : a;
            mssil.put((min - a) / (min > a ? min : a));
        }
    }
    double penalty = 1.;
    // Only if {@link NoiseHandling#IGNORE_NOISE}:
    if (penalize && ignorednoise > 0) {
        penalty = (rel.size() - ignorednoise) / (double) rel.size();
    }
    final double meanssil = penalty * mssil.getMean();
    final double stdssil = penalty * mssil.getSampleStddev();
    if (LOG.isStatistics()) {
        LOG.statistics(new StringStatistic(key + ".simplified-silhouette.noise-handling", noiseOption.toString()));
        if (ignorednoise > 0) {
            LOG.statistics(new LongStatistic(key + ".simplified-silhouette.ignored", ignorednoise));
        }
        LOG.statistics(new DoubleStatistic(key + ".simplified-silhouette.mean", meanssil));
        LOG.statistics(new DoubleStatistic(key + ".simplified-silhouette.stddev", stdssil));
    }
    EvaluationResult ev = EvaluationResult.findOrCreate(db.getHierarchy(), c, "Internal Clustering Evaluation", "internal evaluation");
    MeasurementGroup g = ev.findOrCreateGroup("Distance-based Evaluation");
    g.addMeasure("Simp. Silhouette +-" + FormatUtil.NF2.format(stdssil), meanssil, -1., 1., 0., false);
    db.getHierarchy().resultChanged(ev);
    return meanssil;
}

Example 34 with LongStatistic

use of de.lmu.ifi.dbs.elki.logging.statistics.LongStatistic in project elki by elki-project.

the class RandomProjectedNeighborsAndDensities method computeSetsBounds.

/**
 * Create random projections, project points and put points into sets of size
 * about minSplitSize/2
 *
 * @param points points to process
 * @param minSplitSize minimum size for which a point set is further
 *        partitioned (roughly corresponds to minPts in OPTICS)
 * @param ptList Points that are to be projected
 */
public void computeSetsBounds(Relation<V> points, int minSplitSize, DBIDs ptList) {
    this.minSplitSize = minSplitSize;
    final int size = points.size();
    final int dim = RelationUtil.dimensionality(points);
    this.points = points;
    // perform O(log N+log dim) splits of the entire point sets projections
    int nPointSetSplits = (int) (logOProjectionConst * MathUtil.log2(size * dim + 1));
    // perform O(log N+log dim) projections of the point set onto a random line
    int nProject1d = (int) (logOProjectionConst * MathUtil.log2(size * dim + 1));
    LOG.statistics(new LongStatistic(PREFIX + ".partition-size", nPointSetSplits));
    LOG.statistics(new LongStatistic(PREFIX + ".num-projections", nProject1d));
    splitsets = new ArrayList<>();
    // perform projections of points
    projectedPoints = new DoubleDataStore[nProject1d];
    DoubleDataStore[] tmpPro = new DoubleDataStore[nProject1d];
    Random rand = rnd.getSingleThreadedRandom();
    FiniteProgress projp = LOG.isVerbose() ? new FiniteProgress("Random projections", nProject1d, LOG) : null;
    for (int j = 0; j < nProject1d; j++) {
        double[] currRp = new double[dim];
        double sum = 0;
        for (int i = 0; i < dim; i++) {
            double fl = rand.nextDouble() - 0.5;
            currRp[i] = fl;
            sum += fl * fl;
        }
        sum = FastMath.sqrt(sum);
        for (int i = 0; i < dim; i++) {
            currRp[i] /= sum;
        }
        WritableDoubleDataStore currPro = DataStoreUtil.makeDoubleStorage(ptList, DataStoreFactory.HINT_HOT);
        for (DBIDIter it = ptList.iter(); it.valid(); it.advance()) {
            NumberVector vecPt = points.get(it);
            // Dot product:
            double sum2 = 0;
            for (int i = 0; i < dim; i++) {
                sum2 += currRp[i] * vecPt.doubleValue(i);
            }
            currPro.put(it, sum2);
        }
        projectedPoints[j] = currPro;
        LOG.incrementProcessed(projp);
    }
    LOG.ensureCompleted(projp);
    // Log the number of scalar projections performed.
    long numprod = nProject1d * (long) ptList.size();
    LOG.statistics(new LongStatistic(PREFIX + ".num-scalar-products", numprod));
    // split entire point set, reuse projections by shuffling them
    IntArrayList proind = new IntArrayList(nProject1d);
    for (int j = 0; j < nProject1d; j++) {
        proind.add(j);
    }
    FiniteProgress splitp = LOG.isVerbose() ? new FiniteProgress("Splitting data", nPointSetSplits, LOG) : null;
    for (int avgP = 0; avgP < nPointSetSplits; avgP++) {
        // shuffle projections
        for (int i = 0; i < nProject1d; i++) {
            tmpPro[i] = projectedPoints[i];
        }
        // Shuffle axes (Fisher-Yates)
        for (int i = 1; i < nProject1d; i++) {
            final int j = rand.nextInt(i);
            // Swap i,j
            proind.set(i, proind.set(j, proind.getInt(i)));
        }
        IntIterator it = proind.iterator();
        int i = 0;
        while (it.hasNext()) {
            int cind = it.nextInt();
            projectedPoints[cind] = tmpPro[i];
            i++;
        }
        // split point set
        splitupNoSort(DBIDUtil.newArray(ptList), 0, size, 0, rand);
        LOG.incrementProcessed(splitp);
    }
    LOG.ensureCompleted(splitp);
}

Example 35 with LongStatistic

use of de.lmu.ifi.dbs.elki.logging.statistics.LongStatistic in project elki by elki-project.

the class APRIORI method buildFrequentTwoItemsets.

/**
 * Build the 2-itemsets.
 *
 * @param oneitems Frequent 1-itemsets
 * @param relation Data relation
 * @param dim Maximum dimensionality
 * @param needed Minimum support needed
 * @param ids Objects to process
 * @param survivors Output: objects that had at least two 1-frequent items.
 * @return Frequent 2-itemsets
 */
protected List<SparseItemset> buildFrequentTwoItemsets(List<OneItemset> oneitems, final Relation<BitVector> relation, final int dim, final int needed, DBIDs ids, ArrayModifiableDBIDs survivors) {
    int f1 = 0;
    long[] mask = BitsUtil.zero(dim);
    for (OneItemset supported : oneitems) {
        BitsUtil.setI(mask, supported.item);
        f1++;
    }
    if (LOG.isStatistics()) {
        LOG.statistics(new LongStatistic(STAT + "2-items.candidates", f1 * (long) (f1 - 1)));
    }
    // We quite aggressively size the map, assuming that almost each combination
    // is present somewhere. If this won't fit into memory, we're likely running
    // OOM somewhere later anyway!
    Long2IntOpenHashMap map = new Long2IntOpenHashMap((f1 * (f1 - 1)) >>> 1);
    final long[] scratch = BitsUtil.zero(dim);
    for (DBIDIter iditer = ids.iter(); iditer.valid(); iditer.advance()) {
        BitsUtil.setI(scratch, mask);
        relation.get(iditer).andOnto(scratch);
        int lives = 0;
        for (int i = BitsUtil.nextSetBit(scratch, 0); i >= 0; i = BitsUtil.nextSetBit(scratch, i + 1)) {
            for (int j = BitsUtil.nextSetBit(scratch, i + 1); j >= 0; j = BitsUtil.nextSetBit(scratch, j + 1)) {
                long key = (((long) i) << 32) | j;
                map.put(key, 1 + map.get(key));
                ++lives;
            }
        }
        if (lives > 2) {
            survivors.add(iditer);
        }
    }
    // Generate candidates of length 2.
    List<SparseItemset> frequent = new ArrayList<>(f1 * (int) FastMath.sqrt(f1));
    for (ObjectIterator<Long2IntMap.Entry> iter = map.long2IntEntrySet().fastIterator(); iter.hasNext(); ) {
        Long2IntMap.Entry entry = iter.next();
        if (entry.getIntValue() >= needed) {
            int ii = (int) (entry.getLongKey() >>> 32);
            int ij = (int) (entry.getLongKey() & -1L);
            frequent.add(new SparseItemset(new int[] { ii, ij }, entry.getIntValue()));
        }
    }
    // The hashmap may produce them out of order.
    Collections.sort(frequent);
    if (LOG.isStatistics()) {
        LOG.statistics(new LongStatistic(STAT + "2-items.frequent", frequent.size()));
    }
    return frequent;
}