Two-Mode Overlapping Clustering With Applications to Simultaneous Benefit Segmentation and Market Structuring

Additive Biclustering: A Comparison of One New and Two Existing ALS Algorithms

“…The new algorithm, which incorporates general recommendations regarding optimally designing ALS algorithms, calculates conditionally optimal estimates for the membership matrices and the core matrix of the model in their entirety. In an extensive simulation study, the new algorithm has been shown to outperform two previously proposed algorithms for the same model: P ENCLUS (Both and Gaul 1985) and Clusterwise ALS (Baier et al 1997). As a consequence, the use of the F ull clustering ALS algorithm can be recommended in statistical practice 5 .…”

“…To fit the additive biclustering model in a least squares sense to a data matrix at hand, two alternating least squares (ALS) approaches have been proposed: (1) P ENCLUS Gaul 1987, 1985;Schader and Gaul 1996), and (2) the ALS approach of Baier, Gaul, and Schader (1997), which further will be denoted as Clusterwise ALS. 3 In ALS algorithms, which be-3.…”

A Clustering Model Based on Matrix Approximation with Applications to Cluster System Log Files

“…In this paper, we present a clustering model 1 where the problem of clustering is formulated as matrix approximations. The model explicitly characterizes the data and feature memberships and thus enables the descriptions of each cluster.…”

“…The notations used in the paper are introduced in Table 1. 1 In this paper, we use model and framework interchangeably.…”

“…O(A, X, B) can then be minimized via a two-side iterative procedure (i.e., the natural extensions of the K-means type algorithm for twoside cases [1,3,8].…”