Fuzzy double k-means clustering for simultaneous classification of objects and variables

Two-mode clustering consists in simultaneously clustering modes (e.g., objects, variables) of an observed two-mode data matrix. This idea arises to confront situations in which objects are homogeneous only within subsets of variables, while variables may be strongly associated only on subsets of objects. There are many practical applications presenting the above situations, for example, DNA microarrays analysis and market basket analysis. Other applications include biology, psychology, sociology and so on. We focus on an extension of standard k-means, the double k-means, to simultaneously cluster objects and variables. We propose this model in a fuzzy framework and discuss the advantages of this approach. Finally, we check its adequacy by means of simulation and real case studies.