Visual assessment of co-cluster structure through cooccurrence-sensitive ordering

Visual cluster structure estimation is a useful approach for understanding the intrinsic structural information of data sets without a priori knowledge. Although the VAT algorithm is an efficient heuristic approach with a simple procedure, spectral ordering achieves linearized cluster assignment based on minimization of an objective function. This paper proposes a novel approach for visual co-cluster structure assessment, which is supported by the spectral ordering concept. In a manner similar to the co-clustering extension of the VAT algorithm (coVAT algorithm), object-item cooccurrence information is enlarged into an (object + item) - (object + item) cooccurrence data matrix, and cooccurrence-sensitive ordering is performed through spectral ordering of the enlarged matrix. Additionally, due to the sparse nature of the enlarged matrix, the computational cost can be decreased with the eigen problem of a reduced matrix. The characteristic features of the proposed approach are demonstrated through several numerical experiments.