Detecting Spatial Clusters via a Mixture of Dirichlet Processes

Table 3

Comparison of approaches for selected simulated settings assuming study effect is known.

Setting	Approach	Median Num. of Clusters  (SD)	Cluster index	Average Sensitivity  (SD)	Average Specificity  (SD)	Average Accuracy (%)  (SD)
Normal	DP	3 (0)	1	1 (0)	1 (0)	1 (0)
			2	1 (0)	1 (0)
			3	1 (0)	1 (0)
	Mixture	3 (0)	1	1 (0)	1 (0)	1 (0)
			2	1 (0)	1 (0)
			3	1 (0)	1 (0)
	Kmeans	3 (0.6)	1	0.81 (0.36)	0.93 (0.16)	0.8 (0.22)
			2	0.79 (0.38)	0.94 (0.15)
			3	0.79 (0.25)	1 (0)
Chi-squared	DP	16 (1.76)	1	0.99 (0.01)	0.99 (0.01)	0.65 (0.01)
			2	0.99 (0.01)	0.97 (0.01)
			3	0.14 (0.03)	1 (0)
	Mixture	1 (0.29)	1	0.06 (0.24)	0.98 (0.11)	0.32 (0.1)
			2	0.95 (0.22)	0.06 (0.24)
			3	0.05 (0.21)	1 (0)
	Kmeans	10 (0.22)	1	0.99 (0.06)	0.98 (0.01)	0.62 (0.1)
			2	0.87 (0.34)	0.93 (0.03)
			3	0.16 (0.06)	0.94 (0.17)

SD: standard deviation across 100 MC replicates.