Prediction of organic compound aqueous solubility using machine learning: a comparison study of descriptor-based and fingerprints-based models

Table 4 Gibbs energies and MLR Coefficients for the top twelve features

Substructure	Molecular polarizability	H-bond acceptor	H-bond Donor	∆G²⁹⁸	MLR. coefficients
Feature 1380	12.01	0	0	14.46	−0.93
Feature 807	1.47	0	1	0.29	0.91
Feature 561	10.52	0	0	12.42	−0.54
Feature 222	4.47	1	1	−2.99	0.81
Feature 650	0.80	0	0	−0.90	0.39
Feature 1143	15.02	0	0	18.57	−1.10
Feature 1750	16.69	0	0	20.86	0.01
Feature 1171	2.43	0	0	1.33	0.41
Feature 1873	9.01	0	0	10.34	0.10
Feature 294	17.10	4	0	7.14	−0.013
Feature 114	16.35	0	0	20.40	−0.54
Feature 591	12.68	0	0	15.37	−0.51

ISSN: 1758-2946