A. Svc and . Ward, Figure 1 shows 5 clusters obtained with SVC (right) and HAC (left) Clusters are of arbitrary form, classical clustering algorithms, like k-means, wouldn't be able to find this type of clusters. In order to evaluate the better performing procedure we computed within variance and CATANOVA index (Singh 1993), results are shown in table 1. We obtain a good clustering structure using both methods

S. Abe, Support vector machine for pattern classification, 2005.
DOI : 10.1007/978-1-84996-098-4

P. Lenca, M. Patrick, V. Benò-ot, and L. Stèphane, On selecting interestingness measures for association rules: User oriented description and multiple criteria decision aid, European Journal of Operational Research, vol.184, issue.2, pp.610-626, 2008.
DOI : 10.1016/j.ejor.2006.10.059

K. R. Muller, S. Mika, G. Ratsch, K. Tsuda, and B. Scholkopf, An introduction to kernel-based learning algorithms, IEEE Transactions on Neural Networks, vol.12, issue.2, 2001.
DOI : 10.1109/72.914517

G. Saporta, Simultaneous analysis of qualitative and quantitative data, 1990.

J. Shawe-taylor and N. Cristianini, Kernel methods for pattern analysis, 2004.
DOI : 10.1017/CBO9780511809682

B. Singh, On the Analysis of Variance Method for Nominal Data, The Indian Journal of Statistics, Series B, 1993.