Table 2.

Importance of molecular dynamics (MD) descriptors using correlation-based feature selection (CFS), relief-based attribute selection, and classifier attribute evaluation using artificial neural network (ANN) and random forest (RF) classifiers.

MD descriptor	CFS		Relief-based		Classifier attribute evaluator (ANN)		Classifier attribute evaluator (RF)
	Average merit	Average rank	Average merit	Average rank	Average merit	Average rank	Average merit	Average rank
Total energy	0.062 ± 0.032	2.6 ± 0.92	0.021 ± 0.028	3.3 ± 0.64	−0.015 ± 0.023	2.2 ± 1.47	−0.133 ± 0.032	3.8 ± 0.4
Potential energy	0.062 ± 0.032	1.8 ± 0.75	0.021 ± 0.028	2.3 ± 0.64	−0.015 ± 0.023	2.4 ± 0.49	−0.068 ± 0.046	2.5 ± 0.81
Gyration	0.048 ± 0.033	2.9 ± 1.14	0.014 ± 0.011	3.2 ± 0.98	0 ± 0	2.6 ± 0.49	−0.036 ± 0.024	1.8 ± 0.75
SASA	0.057 ± 0.041	2.7 ± 1.27	0.092 ± 0.013	1.2 ± 0.6	0 ± 0	2.8 ± 1.47	−0.028 ± 0.036	1.9 ± 1.04