stream (version 1.0-3)

DSC_DBSCAN: DBSCAN Macro-clusterer

Description

Implements the DBSCAN algorithm for reclustering micro-clusterings.

Usage

DSC_DBSCAN(eps, MinPts = 5, weighted = TRUE, description=NULL)

Arguments

eps

radius of the eps-neighborhood.

MinPts

minimum number of points required in the eps-neighborhood.

weighted

logical indicating if a weighted version of DBSCAN should be used.

description

optional character string to describe the clustering method.

Value

• An object of class DSC_DBSCAN (a subclass of DSC, DSC_R, DSC_Macro).

Details

DBSCAN is a weighted extended version of the implementation in fpc where each micro-cluster center considered a pseudo point. For weighting we use in the MinPts comparison the sum of weights of the micro-cluster instead of the number.

DBSCAN first finds core points based on the number of other points in its eps-neighborhood. Then core points are joined into clusters using reachability (overlapping eps-neighborhoods).

Note that this clustering cannot be updated iteratively and every time it is used for (re)clustering, the old clustering is deleted.

References

Martin Ester, Hans-Peter Kriegel, Joerg Sander, Xiaowei Xu (1996). A density-based algorithm for discovering clusters in large spatial databases with noise. In Evangelos Simoudis, Jiawei Han, Usama M. Fayyad. Proceedings of the Second International Conference on Knowledge Discovery and Data Mining (KDD-96). AAAI Press. pp. 226-231.

See Also

DSC, DSC_Macro

Examples

# 3 clusters with 5% noise
dsd <- DSD_Gaussians(k=3, d=2, noise=0.05)

# Use DBSCAN to recluster micro clusters (a sample)
dsc <- DSC_Sample(k=100)
cluster(dsc, dsd, 500)

dbscan <- DSC_DBSCAN(eps = .05)
recluster(dbscan, dsc)
plot(dbscan, dsd, type="both")

# For comparison we can cluster some data with DBSCAN directly
# Note: DBSCAN is not suitable for data streams since it passes over the data
# several times.
dbscan <- DSC_DBSCAN(eps = .05)
cluster(dbscan, dsd, 500)
plot(dbscan, dsd)