Examples with QuotientOutlierScoreMeta de.lmu.ifi.dbs.elki.result.outlier.QuotientOutlierScoreMeta used on opensource projects

Example 1 with QuotientOutlierScoreMeta

use of de.lmu.ifi.dbs.elki.result.outlier.QuotientOutlierScoreMeta in project elki by elki-project.

the class LOCI method run.

/**
 * Run the algorithm
 *
 * @param database Database to process
 * @param relation Relation to process
 * @return Outlier result
 */
public OutlierResult run(Database database, Relation<O> relation) {
    DistanceQuery<O> distFunc = database.getDistanceQuery(relation, getDistanceFunction());
    RangeQuery<O> rangeQuery = database.getRangeQuery(distFunc);
    DBIDs ids = relation.getDBIDs();
    // LOCI preprocessing step
    WritableDataStore<DoubleIntArrayList> interestingDistances = DataStoreUtil.makeStorage(relation.getDBIDs(), DataStoreFactory.HINT_TEMP | DataStoreFactory.HINT_SORTED, DoubleIntArrayList.class);
    precomputeInterestingRadii(ids, rangeQuery, interestingDistances);
    // LOCI main step
    FiniteProgress progressLOCI = LOG.isVerbose() ? new FiniteProgress("LOCI scores", relation.size(), LOG) : null;
    WritableDoubleDataStore mdef_norm = DataStoreUtil.makeDoubleStorage(relation.getDBIDs(), DataStoreFactory.HINT_STATIC);
    WritableDoubleDataStore mdef_radius = DataStoreUtil.makeDoubleStorage(relation.getDBIDs(), DataStoreFactory.HINT_STATIC);
    DoubleMinMax minmax = new DoubleMinMax();
    // Shared instance, to save allocations.
    MeanVariance mv_n_r_alpha = new MeanVariance();
    for (DBIDIter iditer = ids.iter(); iditer.valid(); iditer.advance()) {
        final DoubleIntArrayList cdist = interestingDistances.get(iditer);
        final double maxdist = cdist.getDouble(cdist.size() - 1);
        final int maxneig = cdist.getInt(cdist.size() - 1);
        double maxmdefnorm = 0.0;
        double maxnormr = 0;
        if (maxneig >= nmin) {
            // Compute the largest neighborhood we will need.
            DoubleDBIDList maxneighbors = rangeQuery.getRangeForDBID(iditer, maxdist);
            // For any critical distance, compute the normalized MDEF score.
            for (int i = 0, size = cdist.size(); i < size; i++) {
                // Only start when minimum size is fulfilled
                if (cdist.getInt(i) < nmin) {
                    continue;
                }
                final double r = cdist.getDouble(i);
                final double alpha_r = alpha * r;
                // compute n(p_i, \alpha * r) from list (note: alpha_r is not cdist!)
                final int n_alphar = cdist.getInt(cdist.find(alpha_r));
                // compute \hat{n}(p_i, r, \alpha) and the corresponding \simga_{MDEF}
                mv_n_r_alpha.reset();
                for (DoubleDBIDListIter neighbor = maxneighbors.iter(); neighbor.valid(); neighbor.advance()) {
                    // Stop at radius r
                    if (neighbor.doubleValue() > r) {
                        break;
                    }
                    DoubleIntArrayList cdist2 = interestingDistances.get(neighbor);
                    int rn_alphar = cdist2.getInt(cdist2.find(alpha_r));
                    mv_n_r_alpha.put(rn_alphar);
                }
                // We only use the average and standard deviation
                final double nhat_r_alpha = mv_n_r_alpha.getMean();
                final double sigma_nhat_r_alpha = mv_n_r_alpha.getNaiveStddev();
                // Redundant divisions by nhat_r_alpha removed.
                final double mdef = nhat_r_alpha - n_alphar;
                final double sigmamdef = sigma_nhat_r_alpha;
                final double mdefnorm = mdef / sigmamdef;
                if (mdefnorm > maxmdefnorm) {
                    maxmdefnorm = mdefnorm;
                    maxnormr = r;
                }
            }
        } else {
            // FIXME: when nmin was not fulfilled - what is the proper value then?
            maxmdefnorm = Double.POSITIVE_INFINITY;
            maxnormr = maxdist;
        }
        mdef_norm.putDouble(iditer, maxmdefnorm);
        mdef_radius.putDouble(iditer, maxnormr);
        minmax.put(maxmdefnorm);
        LOG.incrementProcessed(progressLOCI);
    }
    LOG.ensureCompleted(progressLOCI);
    DoubleRelation scoreResult = new MaterializedDoubleRelation("LOCI normalized MDEF", "loci-mdef-outlier", mdef_norm, relation.getDBIDs());
    OutlierScoreMeta scoreMeta = new QuotientOutlierScoreMeta(minmax.getMin(), minmax.getMax(), 0.0, Double.POSITIVE_INFINITY, 0.0);
    OutlierResult result = new OutlierResult(scoreMeta, scoreResult);
    result.addChildResult(new MaterializedDoubleRelation("LOCI MDEF Radius", "loci-critical-radius", mdef_radius, relation.getDBIDs()));
    return result;
}

Example 2 with QuotientOutlierScoreMeta

use of de.lmu.ifi.dbs.elki.result.outlier.QuotientOutlierScoreMeta in project elki by elki-project.

the class ALOCI method run.

public OutlierResult run(Database database, Relation<O> relation) {
    final int dim = RelationUtil.dimensionality(relation);
    final Random random = rnd.getSingleThreadedRandom();
    FiniteProgress progressPreproc = LOG.isVerbose() ? new FiniteProgress("Build aLOCI quadtress", g, LOG) : null;
    // Compute extend of dataset.
    double[] min, max;
    {
        double[][] hbbs = RelationUtil.computeMinMax(relation);
        min = hbbs[0];
        max = hbbs[1];
        double maxd = 0;
        for (int i = 0; i < dim; i++) {
            maxd = MathUtil.max(maxd, max[i] - min[i]);
        }
        // Enlarge bounding box to have equal lengths.
        for (int i = 0; i < dim; i++) {
            double diff = (maxd - (max[i] - min[i])) * .5;
            min[i] -= diff;
            max[i] += diff;
        }
    }
    List<ALOCIQuadTree> qts = new ArrayList<>(g);
    double[] nshift = new double[dim];
    ALOCIQuadTree qt = new ALOCIQuadTree(min, max, nshift, nmin, relation);
    qts.add(qt);
    LOG.incrementProcessed(progressPreproc);
    /*
     * create the remaining g-1 shifted QuadTrees. This not clearly described in
     * the paper and therefore implemented in a way that achieves good results
     * with the test data.
     */
    for (int shift = 1; shift < g; shift++) {
        double[] svec = new double[dim];
        for (int i = 0; i < dim; i++) {
            svec[i] = random.nextDouble() * (max[i] - min[i]);
        }
        qt = new ALOCIQuadTree(min, max, svec, nmin, relation);
        qts.add(qt);
        LOG.incrementProcessed(progressPreproc);
    }
    LOG.ensureCompleted(progressPreproc);
    // aLOCI main loop: evaluate
    FiniteProgress progressLOCI = LOG.isVerbose() ? new FiniteProgress("Compute aLOCI scores", relation.size(), LOG) : null;
    WritableDoubleDataStore mdef_norm = DataStoreUtil.makeDoubleStorage(relation.getDBIDs(), DataStoreFactory.HINT_STATIC);
    DoubleMinMax minmax = new DoubleMinMax();
    for (DBIDIter iditer = relation.iterDBIDs(); iditer.valid(); iditer.advance()) {
        final O obj = relation.get(iditer);
        double maxmdefnorm = 0;
        // For each level
        for (int l = 0; ; l++) {
            // Find the closest C_i
            Node ci = null;
            for (int i = 0; i < g; i++) {
                Node ci2 = qts.get(i).findClosestNode(obj, l);
                if (ci2.getLevel() != l) {
                    continue;
                }
                // TODO: always use manhattan?
                if (ci == null || distFunc.distance(ci, obj) > distFunc.distance(ci2, obj)) {
                    ci = ci2;
                }
            }
            // LOG.debug("level:" + (ci != null ? ci.getLevel() : -1) +" l:"+l);
            if (ci == null) {
                // no matching tree for this level.
                break;
            }
            // Find the closest C_j
            Node cj = null;
            for (int i = 0; i < g; i++) {
                Node cj2 = qts.get(i).findClosestNode(ci, l - alpha);
                // TODO: allow higher levels or not?
                if (cj != null && cj2.getLevel() < cj.getLevel()) {
                    continue;
                }
                // TODO: always use manhattan?
                if (cj == null || distFunc.distance(cj, ci) > distFunc.distance(cj2, ci)) {
                    cj = cj2;
                }
            }
            // LOG.debug("level:" + (cj != null ? cj.getLevel() : -1) +" l:"+l);
            if (cj == null) {
                // no matching tree for this level.
                continue;
            }
            double mdefnorm = calculate_MDEF_norm(cj, ci);
            // LOG.warning("level:" + ci.getLevel() + "/" + cj.getLevel() +
            // " mdef: " + mdefnorm);
            maxmdefnorm = MathUtil.max(maxmdefnorm, mdefnorm);
        }
        // Store results
        mdef_norm.putDouble(iditer, maxmdefnorm);
        minmax.put(maxmdefnorm);
        LOG.incrementProcessed(progressLOCI);
    }
    LOG.ensureCompleted(progressLOCI);
    DoubleRelation scoreResult = new MaterializedDoubleRelation("aLOCI normalized MDEF", "aloci-mdef-outlier", mdef_norm, relation.getDBIDs());
    OutlierScoreMeta scoreMeta = new QuotientOutlierScoreMeta(minmax.getMin(), minmax.getMax(), 0.0, Double.POSITIVE_INFINITY);
    OutlierResult result = new OutlierResult(scoreMeta, scoreResult);
    return result;
}

Example 3 with QuotientOutlierScoreMeta

use of de.lmu.ifi.dbs.elki.result.outlier.QuotientOutlierScoreMeta in project elki by elki-project.

the class FlexibleLOF method doRunInTime.

/**
 * Performs the Generalized LOF_SCORE algorithm on the given database and
 * returns a {@link FlexibleLOF.LOFResult} encapsulating information that may
 * be needed by an OnlineLOF algorithm.
 *
 * @param ids Object ids
 * @param kNNRefer the kNN query w.r.t. reference neighborhood distance
 *        function
 * @param kNNReach the kNN query w.r.t. reachability distance function
 * @param stepprog Progress logger
 * @return LOF result
 */
protected LOFResult<O> doRunInTime(DBIDs ids, KNNQuery<O> kNNRefer, KNNQuery<O> kNNReach, StepProgress stepprog) {
    // Assert we got something
    if (kNNRefer == null) {
        throw new AbortException("No kNN queries supported by database for reference neighborhood distance function.");
    }
    if (kNNReach == null) {
        throw new AbortException("No kNN queries supported by database for reachability distance function.");
    }
    // Compute LRDs
    LOG.beginStep(stepprog, 2, "Computing LRDs.");
    WritableDoubleDataStore lrds = DataStoreUtil.makeDoubleStorage(ids, DataStoreFactory.HINT_HOT | DataStoreFactory.HINT_TEMP);
    computeLRDs(kNNReach, ids, lrds);
    // compute LOF_SCORE of each db object
    LOG.beginStep(stepprog, 3, "Computing LOFs.");
    WritableDoubleDataStore lofs = DataStoreUtil.makeDoubleStorage(ids, DataStoreFactory.HINT_STATIC);
    // track the maximum value for normalization.
    DoubleMinMax lofminmax = new DoubleMinMax();
    computeLOFs(kNNRefer, ids, lrds, lofs, lofminmax);
    LOG.setCompleted(stepprog);
    // Build result representation.
    DoubleRelation scoreResult = new MaterializedDoubleRelation("Local Outlier Factor", "lof-outlier", lofs, ids);
    OutlierScoreMeta scoreMeta = new QuotientOutlierScoreMeta(lofminmax.getMin(), lofminmax.getMax(), 0.0, Double.POSITIVE_INFINITY, 1.0);
    OutlierResult result = new OutlierResult(scoreMeta, scoreResult);
    return new LOFResult<>(result, kNNRefer, kNNReach, lrds, lofs);
}

Example 4 with QuotientOutlierScoreMeta

use of de.lmu.ifi.dbs.elki.result.outlier.QuotientOutlierScoreMeta in project elki by elki-project.

the class LDOF method run.

/**
 * Run the algorithm
 *
 * @param database Database to process
 * @param relation Relation to process
 * @return Outlier result
 */
public OutlierResult run(Database database, Relation<O> relation) {
    DistanceQuery<O> distFunc = database.getDistanceQuery(relation, getDistanceFunction());
    KNNQuery<O> knnQuery = database.getKNNQuery(distFunc, k);
    // track the maximum value for normalization
    DoubleMinMax ldofminmax = new DoubleMinMax();
    // compute the ldof values
    WritableDoubleDataStore ldofs = DataStoreUtil.makeDoubleStorage(relation.getDBIDs(), DataStoreFactory.HINT_HOT | DataStoreFactory.HINT_TEMP);
    // compute LOF_SCORE of each db object
    if (LOG.isVerbose()) {
        LOG.verbose("Computing LDOFs");
    }
    FiniteProgress progressLDOFs = LOG.isVerbose() ? new FiniteProgress("LDOF for objects", relation.size(), LOG) : null;
    Mean dxp = new Mean(), Dxp = new Mean();
    for (DBIDIter iditer = relation.iterDBIDs(); iditer.valid(); iditer.advance()) {
        KNNList neighbors = knnQuery.getKNNForDBID(iditer, k);
        dxp.reset();
        Dxp.reset();
        DoubleDBIDListIter neighbor1 = neighbors.iter(), neighbor2 = neighbors.iter();
        for (; neighbor1.valid(); neighbor1.advance()) {
            // skip the point itself
            if (DBIDUtil.equal(neighbor1, iditer)) {
                continue;
            }
            dxp.put(neighbor1.doubleValue());
            for (neighbor2.seek(neighbor1.getOffset() + 1); neighbor2.valid(); neighbor2.advance()) {
                // skip the point itself
                if (DBIDUtil.equal(neighbor2, iditer)) {
                    continue;
                }
                Dxp.put(distFunc.distance(neighbor1, neighbor2));
            }
        }
        double ldof = dxp.getMean() / Dxp.getMean();
        if (Double.isNaN(ldof) || Double.isInfinite(ldof)) {
            ldof = 1.0;
        }
        ldofs.putDouble(iditer, ldof);
        // update maximum
        ldofminmax.put(ldof);
        LOG.incrementProcessed(progressLDOFs);
    }
    LOG.ensureCompleted(progressLDOFs);
    // Build result representation.
    DoubleRelation scoreResult = new MaterializedDoubleRelation("LDOF Outlier Score", "ldof-outlier", ldofs, relation.getDBIDs());
    OutlierScoreMeta scoreMeta = new QuotientOutlierScoreMeta(ldofminmax.getMin(), ldofminmax.getMax(), 0.0, Double.POSITIVE_INFINITY, LDOF_BASELINE);
    return new OutlierResult(scoreMeta, scoreResult);
}

Example 5 with QuotientOutlierScoreMeta

use of de.lmu.ifi.dbs.elki.result.outlier.QuotientOutlierScoreMeta in project elki by elki-project.

the class SimplifiedLOF method run.

/**
 * Run the Simple LOF algorithm.
 *
 * @param database Database to query
 * @param relation Data to process
 * @return LOF outlier result
 */
public OutlierResult run(Database database, Relation<O> relation) {
    StepProgress stepprog = LOG.isVerbose() ? new StepProgress("Simplified LOF", 3) : null;
    DBIDs ids = relation.getDBIDs();
    LOG.beginStep(stepprog, 1, "Materializing neighborhoods w.r.t. distance function.");
    KNNQuery<O> knnq = DatabaseUtil.precomputedKNNQuery(database, relation, getDistanceFunction(), k);
    // Compute LRDs
    LOG.beginStep(stepprog, 2, "Computing densities.");
    WritableDoubleDataStore dens = DataStoreUtil.makeDoubleStorage(ids, DataStoreFactory.HINT_HOT | DataStoreFactory.HINT_TEMP);
    computeSimplifiedLRDs(ids, knnq, dens);
    // compute LOF_SCORE of each db object
    LOG.beginStep(stepprog, 3, "Computing SLOFs.");
    WritableDoubleDataStore lofs = DataStoreUtil.makeDoubleStorage(ids, DataStoreFactory.HINT_STATIC);
    DoubleMinMax lofminmax = new DoubleMinMax();
    computeSimplifiedLOFs(ids, knnq, dens, lofs, lofminmax);
    LOG.setCompleted(stepprog);
    // Build result representation.
    DoubleRelation scoreResult = new MaterializedDoubleRelation("Simplified Local Outlier Factor", "simplified-lof-outlier", lofs, ids);
    OutlierScoreMeta scoreMeta = new QuotientOutlierScoreMeta(lofminmax.getMin(), lofminmax.getMax(), 0., Double.POSITIVE_INFINITY, 1.);
    OutlierResult result = new OutlierResult(scoreMeta, scoreResult);
    return result;
}