Table 5.

Results of the second simulation study. The following methods were applied to the simulated data set described in Section 2.4: (1) supervised sparse clustering, (2) sparse clustering, (3) supervised clustering (Bair and Tibshirani, 2004), (4) k-means clustering on the top 3 principal component (PCA) scores. The mean number of misclassified observations (and associated standard errors) is shown for each method.

	Sup. sparse clust.	Sparse clust.	Sup. clust.	Clust. on PCA
Mean	7.2	94.5	11.3	95.1
SE	0.8	0.1	0.7	0.1