Agglomerative hierarchical clustering of continuous variables based on mutual information

摘要

In order to study interdependencies among continuous variables in the framework of a data analysis problem, an agglomerative hierarchical clustering of the set of variables is performed. The similarity measure used within the clustering algorithm is based on the notion of mutual information. Recent results on the estimation of this measure of stochastic dependence are presented and the behavior of the clustering algorithm is studied on several artificial problems, i.e., which "structure" is known.