Harnessing Shannon entropy-based descriptors in machine learning models to enhance the prediction accuracy of molecular properties

Table 3 Comparison of Morgan, SEF, SHED and hybrid descriptors in random forest regression-based models

Dataset (Target ID)	Target variable	Morgan (MAE)	SEF (MAE)	SHED (MAE)	SEF + Morgan (MAE)	SEF + SHED (MAE)
CHEMBL 3713062^a	BEI	4.75 ± 0.01	2.92 ± 0.03	4.51 ± 0.02	2.36 ± 0.02	2.92 ± 0.03
CHEMBL 204	BEI	4.60 ± 0.05	3.65 ± 0.03	7.10 ± 0.05	3.23 ± 0.00	3.60 ± 0.00
CHEMBL 2842	BEI	4.10 ± 0.02	3.61 ± 0.00	8.35 ± 0.05	3.13 ± 0.01	3.60 ± 0.01
CHEMBL 274	BEI	2.20 ± 0.04	2.10 ± 0.01	3.85 ± 0.01	1.85 ± 0.00	2.10 ± 0.00
CHEMBL 5023^b	logP	0.51 ± 0.00	0.43 ± 0.01	0.63 ± 0.01	0.43 ± 0.00	0.40 ± 0.00

(i) ^aThe scaling factor of MAE was 10⁵, ^bthe scaling factor was 1 and for the rest of the Target IDs the scaling factor was 10³ and (ii) ^bMAE with the kNN-based model was 0.50

ISSN: 1758-2946