Table 2.

Top ranked features

Feature ranking method
Rank	Anova (mean ± sd)	IG (mean ± sd)	RF (mean ± sd)
1	ID (8.78 ±1.27)	ID (0.44 ± 0.05)	ID (0.18 ± 0.01)
2	C (7.29 ± 0.82)	C (0.08 ± 0.14)	C (0.14 ± 0.01)
3	IC,R (5.96 ± 0.68)	IC,R (0.05 ± 0.12)	J (0.11 ± 0.01)
4	J (4.12 ± 0.58)	$I_{f_{\mathsf{\text{exp}}}^P}^{\lambda }$ (0.02 ± 0.08)	$I_{f_{lin}^V}$ (0.11 ± 0.01)
5	$I_{f_{\mathsf{\text{exp}}}^P}^{\lambda }$ (4.04 ± 0.65)	$I_{I_{lin}^C}^{\lambda }$ (0.01 ± 0.04)	$I_{f_{exp}^V}$ (0.105 ± 0.01)

This table shows the five top ranked features using different feature selection methods. A description of the different descriptors (features) can be found in the additional file.