The development of models to predict melting and pyrolysis point data associated with several hundred thousand compounds mined from PATENTS

Table 1 The number of compounds and average properties of molecules of the analyzed datasets and their drug-like subsets

Dataset	Type	Whole set				Drug-like set, % of the total set
		N	Average
		N	T (°C)	MW	NA
PATENTS	Training	241,958	159	357	25	89
Decomposing	Training	13,785	209	358	25	76
Non-decomposing	Training	228,173	155	357	25	93
Bergström	Validation	277	151	295	20.8	92
Bradley	Validation	2878	59	174	11.4	53
OCHEM	Validation	21,832	117	249	16.7	73
Enamine	Validation	22,449	143	223	14.9	91
COMBINED	Validation, merge of four sets	47,436	126	233	15.6	81

ISSN: 1758-2946