A New Simultaneous Two-Levels Coclustering Algorithm for Behavioural Data-Mining

Clustering is a very powerful tool for automatic detection of relevant sub-groups in unlabeled data sets. It can be sometime very interesting to be able to regroup and visualize the attributes used to describe the data, in addition to the clustering of these data. In this paper, we propose a coclustering algorithm based on the learning of a Self Organizing Map. The new algorithm will thus be able at the same time to map data and features in a low dimensional sub-space, allowing simple visualization, and to produce a clustering of both data and features. The resulting output is therefore very informative and easy to analyze.