Cross-validation pitfalls when selecting and assessing regression and classification models

Table 3 Selected optimal cross-validatory chosen models

Dataset	Model	Lowest average cross_validation loss	Optimal parameter	Min grid value	Max grid value	Grid size
AquaticTox	PLS	0.5948	13	1	60	60
AquaticTox	Ridge	0.5767	0.05325	0.00107	1069.93	100
bbb2	Ridge	0.1689	0.10494	0.00026	260	100
Caco-PipelinePilotFP	Ridge	0.0916	0.008058	0.00025	246	100
Caco-QuickProp	Ridge	0.1162	0.0279	0.00021	211	100
MeltingPoint	PLS	45.5848	47	1	60	60
MeltingPoint	Ridge	45.4370	0.0549	0.02734	27346.2	100
Mutagen	Ridge	0.1889	0.003142	0.00017	168	100
PLD	Ridge	0.1768	1.02431	0.00021	205.81	100

ISSN: 1758-2946