Evolutionary Algorithms for Robust Density-Based Data Clustering

Table 5

Clustering results for contaminated Iris datasets.

Iris dataset with n = 165, 10% noise
	2GA	DBSCAN-RC
Generations till convergence, G	51	32
% individuals with c = 2	16	10
% individuals with c = 3	84	90
Average specificity (family of c = 2)	80.00	93.33
Average specificity (family of c = 3)	86.67	93.33
Average sensitivity (family of c = 2)	88.00	91.33
Average sensitivity (family of c = 3)	94.67	97.33
Iris dataset with n = 180, 20% noise
	2GA	DBSCAN-RC
Generations till convergence, G	57	36
% individuals with c = 2	25	10
% individuals with c = 3	67	88
Average specificity (family of c = 2)	83.33	90.00
Average specificity (family of c = 3)	90.00	93.33
Average sensitivity (family of c = 2)	86.67	90.66
Average sensitivity (family of c = 3)	93.33	96.66
Iris dataset with n = 195, 30% noise
	2GA	DBSCAN-RC
Generations till convergence, G	62	37
% individuals with c = 2	33	12
% individuals with c = 3	67	88
Average specificity (family of c = 2)	73.33	91.11
Average specificity (family of c = 3)	77.77	95.55
Average sensitivity (family of c = 2)	85.33	90.66
Average sensitivity (family of c = 3)	88.00	92.66