Feature Selection for Better Identification of Subtypes of Guillain-Barré Syndrome

Table 6

Purity for different numbers of clusters using the four GBS subtypes.

	IG	SU	CFS	Consistency	Chi-squared	All features
2	0.6434 (9)	0.6434 (10)	0.6511 (16)	0.4883 (6)	0.6124 (48)	0.5038
3	0.7906 (7)	0.7906 (7)	0.7286 (16)	0.6666 (6)	0.7829 (6)	0.5813
4	0.7984 (7)	0.7984 (7)	0.7984 (16)	0.6589 (6)	0.7829 (41)	0.6899
5	0.7984 (5)	0.7984 (5)	0.7906 (16)	0.7286 (6)	0.7829 (91)	0.6821
6	0.7751 (4)	0.7751 (7)	0.8139 (16)	0.7596 (6)	0.7751 (38)	0.6666
10	0.8139 (5)	0.8139 (5)	0.8217 (16)	0.7596 (6)	0.8062 (38)	0.6976
20	0.8449 (38)	0.8372 (31)	0.8527 (16)	0.8294 (6)	0.8294 (53)	0.7596

: number of clusters, IG: information gain, and SU: symmetrical uncertainty. The number of features selected in each case is shown in parenthesis.