Table 2:

Results of classification under various model assumptions to recover a simulated 4 clusters of 60 each. Down the rows we have an increasing α, which should encourage the creation of new clusters. Across the columns from left to right we have an increasing measurement error of ${\sigma }_y^2$, which should discourage the creation of new clusters.

		${\sigma }_y^2=0.50$					${\sigma }_y^2=1.00$				${\sigma }_y^2=1.50$				${\sigma }_y^2=3.00$
α = 0.01	ci	1	2	3	4	5	1	2	3	4	1	2	3		1
	nk	64	33	63	60	20	79	67	35	59	145	30	65		240
α = 1.00	ci	1	2	3	4	5	1	2	3	4	1	2	3	4	1
	nk	65	60	30	22	63	60	36	68	76	87	30	66	57	240
α = 3.00	ci	1	2	3	4	5	1	2	3	4	1	2	3	4	1
	nk	66	60	24	27	63	68	60	34	78	36	63	74	37	240
α = 5.00	ci	1	2	4	5		1	2	3	4	1	3	5	6	1
	nk	75	61	33	71		84	61	64	31	33	79	113	15	240