A Proposal of an Extended Model of ADCLUS Model
This paper presents an extended model of ADCLUS model as a model for overlapping clustering. It is assumed that a degree of the belongings of each object to a cluster is represented as a binary random variable, and similarity between two objects is defined by a weighted expectation of a cross product of these variables. The problems on the visualization of the results is also discussed. An extension of the proposed model to a two-mode, two-way data is also described.
KeywordsExtended Model Space Structure Latent Class Model Binary Random Variable Morse Code
Unable to display preview. Download preview PDF.
- Carroll, J. D. and Arabie, P. (1983): LNDCLUS: An individual differences generalization of the ADCLUS model and the MAPCLUS algorithm Psychometrika, 48, 157–169.Google Scholar
- Shepard, R.N., and Arabie, P. (1979): Additive Clustering Representation of Similaritiesas as Combinations of Descrete Overlapping Properties. Psychological Review, 86, 87–123.Google Scholar