This function allows you to combine a micro clustering algorithm and a macro
clustering algorithm into a single object that can then be used as a normal
DSC object. This object can then be used with e.g. update to produce a
micro clustering of a stream.
Usage
DSC_ThreeStage(macro, micro)
Arguments
macro
a dsc object representing a macroclustering
micro
a dsc object representing a micro clustering, which can be
obtained by calling e.g. DSC_subspaceCluStream.
Details
The microclustering component is implemented as usual, i.e. a
DSC_ThreeStage object with a DSC_subspaceCluStream
microclustering component will produce the same microclusters that a normal
CluStream would have produced. This is the first of the three stages. Whenever a macro clustering is requested, Multivariate Gaussian Distributions
around the positions of the microclusters are used to simulate the original
stream. This is the second stage. Then the macro clustering is performed on the points generated by these
distributions using the selected macro clustering algorithm. This is the third
stage. Possible choices for the micro clusterer are DSC_subspaceDenStream and
DSC_subspaceCluStream. Possible macro clusterers are DSC_clique,
DSC_p3c, DSC_proclus and DSC_subclu. Other clusterers are
currently not supported.