B1133
Title: On the interpretation of multiple clusterings
Authors: Ryan Browne - University of Waterloo (Canada) [presenting]
Abstract: We explore the idea of identifying different partitions based on mutually exclusive sets of variables with a dataset. In the literature this concept is known as multiple clusterings or multiple cluster structures. We suggest this is a misnomer as the methodology yields a single clustering solution. However, these types of solutions, which we call independent clusters, give rise to different interpretations of clusterings then the usual several distinct components. We relax the assumption that these clusters are composed of variables subsets to directional subsets. Along the way we demonstrate this methodology on simulated and real data.
