A hybrid framework for improving uncertainty quantification in deep learning-based QSAR regression modeling

Table 1 Average performance of four individual models, BYS and NLLCAL

Metrics	Splitting strategies	MVE	ENS	LDIST	FDIST	BYS	NLLCAL
SCC	IVIT	0.212	0.257	0.281	0.161	0.263	0.308
	IVOT	0.154	0.193	0.202	0.134	0.198	0.225
	OVOT	0.171	0.142	0.174	0.101	0.183	0.194
ECE	IVIT	0.198	0.310	NA^a	NA	0.151	0.030
	IVOT	0.295	0.344	NA	NA	0.243	0.058
	OVOT	0.149	0.332	NA	NA	0.121	0.059
ENCE	IVIT	1.576	2.903	NA	NA	0.943	0.184
	IVOT	2.849	3.582	NA	NA	1.656	0.287
	OVOT	0.773	3.575	NA	NA	0.572	0.258

ISSN: 1758-2946