Table 3. Performance summary of QSAR models assessed using additional statistical metrics.

Descriptor class	N	Training set		10-fold CV set		External set		iR2	iQ2
		$r_m^2$	$\mathrm{\Delta }{r}_m^2$	$r_m^2$	$\mathrm{\Delta }{r}_m^2$	$r_m^2$	$\mathrm{\Delta }{r}_m^2$
CDK	960	0.82 ± 0.02	0.07 ± 0.01	0.62 ± 0.09	0.20 ± 0.06	0.64 ± 0.05	0.19 ± 0.03	0.0003	−0.0003
CDK extended	948	0.83 ± 0.01	0.07 ± 0.01	0.62 ± 0.08	0.20 ± 0.05	0.65 ± 0.05	0.18 ± 0.03	0.0006	−0.0005
CDK graph only	198	0.70 ± 0.02	0.14 ± 0.01	0.51 ± 0.07	0.27 ± 0.05	0.53 ± 0.05	0.26 ± 0.03	0.0007	−0.0006
E-State	21	0.35 ± 0.03	0.38 ± 0.02	0.27 ± 0.07	0.40 ± 0.04	0.28 ± 0.05	0.41 ± 0.03	0.0011	−0.0009
MACCS	77	0.73 ± 0.01	0.12 ± 0.01	0.57 ± 0.09	0.23 ± 0.05	0.58 ± 0.05	0.23 ± 0.03	0.0005	−0.0004
PubChem	103	0.74 ± 0.02	0.12 ± 0.01	0.57 ± 0.07	0.23 ± 0.04	0.59 ± 0.05	0.22 ± 0.03	0.0006	−0.0005
Substructure	30	0.50 ± 0.02	0.28 ± 0.01	0.39 ± 0.07	0.34 ± 0.05	0.41 ± 0.05	0.33 ± 0.03	0.0033	–0.0027
Substructure count	26	0.77 ± 0.01	0.10 ± 0.01	0.60 ± 0.08	0.22 ± 0.05	0.61 ± 0.06	0.21 ± 0.04	0.0015	−0.0013
Klekota–Roth	111	0.71 ± 0.02	0.14 ± 0.01	0.54 ± 0.08	0.25 ± 0.05	0.56 ± 0.06	0.24 ± 0.03	0.0006	–0.0004
Klekota–Roth count	72	0.76 ± 0.02	0.11 ± 0.01	0.60 ± 0.08	0.22 ± 0.05	0.61 ± 0.07	0.21 ± 0.04	0.0006	−0.0005
2D atom pairs	42	0.49 ± 0.03	0.28 ± 0.02	0.35 ± 0.08	0.36 ± 0.04	0.35 ± 0.06	0.36 ± 0.03	0.0010	–0.0008
2D atom pairs count	38	0.75 ± 0.01	0.10 ± 0.01	0.52 ± 0.05	0.26 ± 0.05	0.54 ± 0.06	0.25 ± 0.04	0.0006	−0.0005