A new topological clustering algorithm for interval data

• We propose a clustering algorithm for interval data, based on a Self-Organizing Map.
• Its major advantage is that the number of clusters to find is detected automatically.
• The results confirm the effectiveness of the algorithm to deal with interval data.
• The algorithm discriminates perfectly overlapped groups of different shapes.

Clustering is a very powerful tool for automatic detection of relevant sub-groups in unlabeled data sets. In this paper we focus on interval data: i.e., where the objects are defined as hyper-rectangles. We propose here a new clustering algorithm for interval data, based on the learning of a Self-Organizing Map. The major advantage of our approach is that the number of clusters to find is determined automatically; no a priori hypothesis for the number of clusters is required. Experimental results confirm the effectiveness of the proposed algorithm when applied to interval data.