Spatial Data Clustering using Tolerance Relation

F.-S. Sun and C.-H. Tzeng (USA)


Data Clustering, Data Mining, Machine Learning.


This paper introduces a novel spatial data clustering method based on tolerance relation. Two specific clusters, representative and closure clusters, are discussed. A rep resentative clustering in general is not a partition, and thus is suitable for spatial clustering, in which overlapped areas are commonly present. The closure clustering could iden tify clusters with arbitrary shapes naturally. We demon strate the effectiveness of this method by experiments us ing synthetic data and show its comparison with -means algorithm.

Important Links:

Go Back