Table 2.

Statistical metrics (root-mean-square error RMSE, mean absolute error MAE, mean signed error MSE, and slope m′, intercept b′, and coefficient of determination R² from descriptive regression) for all compounds from SAMPL6-type models for water [MP2/6-311+G(d,p)/PSE-2] and cyclohexane (PSE-2) and the original SAMPL5 setup

Setup	Observable	Cyclohexane mod	Batches	RMSE	MSE	MAE	R2	m′	b′
SAMPL6	log P	1-par	0 + 1 + 2	3.40	1.25	2.59	0.52	1.76	1.60
		2-par	0 + 1 + 2	4.36	3.38	3.67	0.56	1.65	3.69
		2-par-I	0 + 1 + 2	2.33	− 0.01	1.76	0.54	1.4	0.18
		3-par	0 + 1 + 2	3.18	2.21	2.68	0.57	1.45	2.42
	log D7.4	1-par	0 + 1 + 2	3.23	0.59	2.50	0.63	2.02	1.07
		2-par	0 + 1 + 2	3.97	2.72	3.40	0.65	1.92	3.15
		2-par-I	0 + 1 + 2	2.46	− 0.67	1.71	0.67	1.69	− 0.35
		3-par	0 + 1 + 2	2.88	1.55	2.44	0.66	1.72	1.89
SAMPL5	log P	2-par-I(5)	0 + 1 + 2	2.33	0.55	1.79	0.55	1.39	0.73
		3-par(5)	0 + 1 + 2	3.63	2.85	3.10	0.57	1.43	3.04
	log D7.4	2-par-I(5)	0 + 1 + 2	2.32	− 0.37	1.76	0.68	1.69	− 0.05
		3-par(5)	0 + 1 + 2	3.11	1.92	2.74	0.66	1.73	2.27

Optimized solution structures are provided as Online Resource 3; calculated data, also split into separate components, as Online Resource 4