Class FDBSCANNeighborPredicate

  • All Implemented Interfaces:

    @Reference(authors="Hans-Peter Kriegel, Martin Pfeifle",
               title="Density-based clustering of uncertain data",
               booktitle="Proc. 11th ACM Int. Conf. on Knowledge Discovery and Data Mining (SIGKDD)",
    public class FDBSCANNeighborPredicate
    extends java.lang.Object
    implements NeighborPredicate<DBIDs>
    Density-based Clustering of Applications with Noise and Fuzzy objects (FDBSCAN) is an Algorithm to find sets in a fuzzy database that are density-connected with minimum probability.


    Hans-Peter Kriegel, Martin Pfeifle
    Density-based clustering of uncertain data
    Proc. 11th ACM Int. Conf. on Knowledge Discovery and Data Mining (SIGKDD)

    This class is a NeighborPredicate presenting this Algorithm in use with GeneralizedDBSCAN.

    Only Euclidean distance is supported, because of the pruning strategy described in the original article which needs minimum and maximum distances of bounding rectangles. Index support is not yet available.

    Alexander Koos, Erich Schubert
    • Field Detail

      • epsilon

        protected double epsilon
        Epsilon radius
      • sampleSize

        protected int sampleSize
        The size of samplesets that should be drawn for neighborcheck.
      • threshold

        protected double threshold
        The relative amount of epsilon-close pairings determined by the neighborcheck.
      • rand

        protected RandomFactory rand
        The random generator to draw the samples with.
    • Constructor Detail

      • FDBSCANNeighborPredicate

        public FDBSCANNeighborPredicate​(double epsilon,
                                        int sampleSize,
                                        double threshold,
                                        RandomFactory seed)
        epsilon - Maximum distance
        sampleSize - Sampling size
        threshold - Threshold on how many samples are within the radius
        seed - Random generator for sampling