Table IV.

Performance of the Best PLS Models for the Seven Drugs

Model	Drug	Interaction fields	n1a	NPCb	Nc	r2 d	q2 e	RMSEf	XRMESg	n2h	$r_{\text{test}}^2$ i	RMSEtestj
1	APV	ΔEvdw, ΔGpolar	1551	50	118	0.852	0.814	0.573	0.645	776	0.810	0.636
2	ATV	ΔEvdw, ΔEele	531	110	125	0.898	0.805	0.514	0.719	265	0.828	0.669
3	IDV	ΔEvdw, ΔGpolar	1629	50	124	0.870	0.838	0.620	0.694	815	0.842	0.657
4	LPV	ΔEvdw, ΔEele, ΔGGB	1074	95	214	0.909	0.864	0.580	0.712	537	0.885	0.631
5	NFV	ΔEvdw, ΔEele, ΔGGB	1643	100	220	0.899	0.871	0.607	0.687	821	0.859	0.712
6	RTV	ΔEvdw, ΔEele, ΔGGB	1605	120	223	0.940	0.908	0.513	0.635	802	0.921	0.581
7	SQV	ΔEvdw, ΔEele, ΔGGB	1630	120	223	0.901	0.839	0.623	0.802	815	0.855	0.730

^an₁ is the number of samples in the training set.

^bN_PC is the number of principal components used in the model.

^cN is the number of MIECs terms.

^dr² is the squared regression coefficient for the training set.

^eq² is the squared leave-one-out cross-validation regression coefficient.

^fRMSE is the root mean square error for the training set.

^gXRMSE is the leave-one-out cross-validation root mean square error for the training set.

^hn₂ is the number of samples in the test set.

ⁱ $r_{\text{test}}^2$ is the squared regression coefficient for the test set.

^jRMSE_test is the root mean square error for the test set.