Table 1. Optimisation and performance statistics of the QSPR models developed for the training set.

Method	Resampling results of optimal model	Performance of optimal model on training set
	H measure ± SD	Kappa	Balanced accuracy	P value	H measure	Overall quality of model
SVM	0.764 ± 0.28	0.941	0.971	2.04 × 10–9	1
RF	0.771 ± 0.22	0.941	0.971	2.04 × 10–9	1
kNN	0.570 ± 0.26	0.824	0.912	3.83 × 10–7	0.738
NN	0.774 ± 0.24	0.941	0.971	2.04 × 10–9	0.907
PLS	0.751 ± 0.22	0.529	0.765	1.47 × 10–3	0.761
NB	0.701 ± 0.24	0.765	0.882	3.08 × 10–6	0.761
C5.0	0.646 ± 0.25	1	1	5.82 × 10–11	1