Package elki.outlier.spatial

Class SLOM<N,​O>

  • Type Parameters:
    N - the type the spatial neighborhood is defined over
    O - the type of objects handled by the algorithm
    All Implemented Interfaces:
    elki.Algorithm, OutlierAlgorithm
    @Title("SLOM: a new measure for local spatial outliers")
@Description("Spatial local outlier measure (SLOM), which captures the local behaviour of datum in their spatial neighbourhood")
@Reference(authors="S. Chawla, P. Sun",
           title="SLOM: a new measure for local spatial outliers",
           booktitle="Knowledge and Information Systems 9(4)",
           url="https://doi.org/10.1007/s10115-005-0200-2",
           bibkey="DBLP:journals/kais/ChawlaS06")
public class SLOM<N,​O>
extends AbstractDistanceBasedSpatialOutlier<N,​O>
    SLOM: a new measure for local spatial outliers

    Reference:

    S. Chawla, P. Sun
    SLOM: a new measure for local spatial outliers
    Knowledge and Information Systems 9(4)

    This implementation works around some corner cases in SLOM, in particular when an object has none or a single neighbor only (albeit the results will still not be too useful then), which will result in divisions by zero.

    Since:
    0.4.0
    Author:
    Ahmed Hettab

    • Constructor Detail

      • SLOM

        public SLOM​(NeighborSetPredicate.Factory<N> npred,
            elki.distance.PrimitiveDistance<O> nonSpatialDistance)
        Constructor.
        Parameters:
        npred - Neighborhood predicate
        nonSpatialDistance - Distance function to use on the non-spatial attributes

    • Method Detail

      • getInputTypeRestriction

        public elki.data.type.TypeInformation[] getInputTypeRestriction()

      • run

        public OutlierResult run​(elki.database.Database database,
                         elki.database.relation.Relation<N> spatial,
                         elki.database.relation.Relation<O> relation)
        Parameters:
        database - Database to process
        spatial - Spatial Relation to use.
        relation - Relation to use.
        Returns:
        Outlier detection result