P3C: A Robust Projected Clustering Algorithm

摘要: Projected clustering has emerged as a possible solution to the challenges associated with in high dimensional data. A projected cluster is subset of points together attributes, such that project onto small range values each these and are uniformly distributed remaining attributes. Existing algorithms for rely on parameters whose appropriate difficult set by user, or unable identify clusters few relevant In this paper, we present robust algorithm can effectively discover data while minimizing number required input. contrast all previous approaches, our discover, under very general conditions, true clusters. We show through an extensive experimental evaluation algorithm: (1) significantly outperforms existing terms accuracy; (2) effective detecting low-dimensional embedded spaces; (3) varying orientation their subspaces; (4) scalable respect large sets dimensions.