Table 1. Training and Testing Scores for QSAR Models Tested on Different Parameters of Interest^a.

	Data set MSE
Model	AutoDock Vina train	AutoDock Vina test	CNS MPO train	CNS MPO test	Aggregation train	Aggregation test
LR	0.699	0.878	0.366	0.416	0.0002	O(1011)
DT	0.009	0.359	0	0.487	0.0002	0.122
RF	0.047	0.178	0.28	0.232	0.052	0.516
DNN	0.304	0.415	0.056	0.123	0.017	0.081

^aAverage mean square error (MSE) from cross validation for four different models for the AutoDock Vina scores and CNS MPO scores (training and testing on the Cayman dataset of ∼10,000 compounds) and half-times of aggregation (training and testing on the aggregation dataset of ∼300 compounds¹⁸⁻²¹). LR = linear regressor, DT = decision tree, RF = random forest, and DNN = deep neural network.