subspaceMOA (version 0.6.0)

DSC_PreDeConStream: Density-Based Projected Clustering of Data Streams

Description

This function creates a DSC object that represents an instance of the PreDeConStream algorithm and can be used for stream clustering.

Usage

DSC_PreDeConStream(epsilonN = 0.7, beta = 0.3, muN = 10, muF = 3,
  lambda = 0.1, initPoints = 1000, tau = 2, kappa = 10, delta = 0.01,
  offline = 2, speed = 100)

Arguments

epsilonN

radius of each neighborhood

beta

control the effect of mu

muN

minimum number of points in microclusters

muF

minimum number of points in macroclusters

lambda

decaying parameter

initPoints

number of points to use for initialization

tau

number of maximal subspace dimensionality

kappa

parameter to define preference weighted vector

delta

defines the threshold for the variance

offline

offline multiplier for epsilon

speed

processing number of incoming points per time unit

Details

The PreDeConStream algorithm is a Density-Based algorithm for the projected clustering of data streams. To initially obtain a set of microclusters initPoints points are buffered and clustered using the PreDeCon algorithm. Then, microclusters are maintained by checking for each new point whether it falls within the radius of an existing microcluster, similar to DSC_DenStream. Microclusters are aged according to a decay paramter lambda. Macroclusters are also maintained throughout the run of the algorithm by updating the affected macroclusters, whenever a change in the microcluster structure has occured, using a component of the PreDeCon algorithm to do so.

Examples

dsc <- DSC_PreDeConStream()
dsd <- DSD_RandomRBFSubspaceGeneratorEvents()
update(dsc,dsd,1000)