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

Table 5

Purity of pairwise clustering of GBS subtypes.

GBS subtypes	IG	SU	CFS	Consistency	Chi-squared	All features
AIDP and AMAN	0.9649 (2)	0.9649 (2)	*	0.9824 (2)	0.9649 (2)	0.8771
AMAN and AMSAN	0.927 (3)	0.9375 (2)	0.875 (7)	0.927 (3)	0.9687 (4)	0.7395
AIDP and AMSAN	0.962 (3)	0.9367 (2)	0.9113 (9)	0.7468 (5)	0.962 (3)	0.8354
AIDP and MF	0.8787 (3)	0.8787 (4)	0.8787 (4)	0.8787 (3)	0.909 (3)	0.6666
AMAN and MF	0.98 (6)	0.96 (3)	0.98 (4)	0.74 (2)	0.98 (6)	0.96
AMSAN and MF	0.9305 (13)	0.9444 (13)	0.9583 (14)	0.8194 (5)	0.9444 (14)	0.8333

IG: information gain, SU: symmetrical uncertainty, and *one feature selected therefore purity was not computed. The number of features selected in each case is shown in parenthesis.