An Optimal centroid based actionable 3D subspace clustering

Abstract

An efficient Optimal Centroid-based Actionable 3D Subspace Clustering from continuous valued 3D data which is suitable for real world problems like discovery of profitable stocks, biologically significant protein residues etc. lt achieves actionability(ability to suggest profitable or beneficial actions to users), incorporation of domain knowledge which allows users to choose the preferred utility function, parameter insensitivity and excellent performance through a set of centroids generetaed by Single Pass Seed Selection and by the combination of singular value decomposition, numerical optimization and 3D closed frequent item set mining.